Search Results

Search found 6612 results on 265 pages for 'seconds'.

Page 62/265 | < Previous Page | 58 59 60 61 62 63 64 65 66 67 68 69  | Next Page >

  • Heaps of Trouble?

    - by Paul White NZ
    If you’re not already a regular reader of Brad Schulz’s blog, you’re missing out on some great material.  In his latest entry, he is tasked with optimizing a query run against tables that have no indexes at all.  The problem is, predictably, that performance is not very good.  The catch is that we are not allowed to create any indexes (or even new statistics) as part of our optimization efforts. In this post, I’m going to look at the problem from a slightly different angle, and present an alternative solution to the one Brad found.  Inevitably, there’s going to be some overlap between our entries, and while you don’t necessarily need to read Brad’s post before this one, I do strongly recommend that you read it at some stage; he covers some important points that I won’t cover again here. The Example We’ll use data from the AdventureWorks database, copied to temporary unindexed tables.  A script to create these structures is shown below: CREATE TABLE #Custs ( CustomerID INTEGER NOT NULL, TerritoryID INTEGER NULL, CustomerType NCHAR(1) COLLATE SQL_Latin1_General_CP1_CI_AI NOT NULL, ); GO CREATE TABLE #Prods ( ProductMainID INTEGER NOT NULL, ProductSubID INTEGER NOT NULL, ProductSubSubID INTEGER NOT NULL, Name NVARCHAR(50) COLLATE SQL_Latin1_General_CP1_CI_AI NOT NULL, ); GO CREATE TABLE #OrdHeader ( SalesOrderID INTEGER NOT NULL, OrderDate DATETIME NOT NULL, SalesOrderNumber NVARCHAR(25) COLLATE SQL_Latin1_General_CP1_CI_AI NOT NULL, CustomerID INTEGER NOT NULL, ); GO CREATE TABLE #OrdDetail ( SalesOrderID INTEGER NOT NULL, OrderQty SMALLINT NOT NULL, LineTotal NUMERIC(38,6) NOT NULL, ProductMainID INTEGER NOT NULL, ProductSubID INTEGER NOT NULL, ProductSubSubID INTEGER NOT NULL, ); GO INSERT #Custs ( CustomerID, TerritoryID, CustomerType ) SELECT C.CustomerID, C.TerritoryID, C.CustomerType FROM AdventureWorks.Sales.Customer C WITH (TABLOCK); GO INSERT #Prods ( ProductMainID, ProductSubID, ProductSubSubID, Name ) SELECT P.ProductID, P.ProductID, P.ProductID, P.Name FROM AdventureWorks.Production.Product P WITH (TABLOCK); GO INSERT #OrdHeader ( SalesOrderID, OrderDate, SalesOrderNumber, CustomerID ) SELECT H.SalesOrderID, H.OrderDate, H.SalesOrderNumber, H.CustomerID FROM AdventureWorks.Sales.SalesOrderHeader H WITH (TABLOCK); GO INSERT #OrdDetail ( SalesOrderID, OrderQty, LineTotal, ProductMainID, ProductSubID, ProductSubSubID ) SELECT D.SalesOrderID, D.OrderQty, D.LineTotal, D.ProductID, D.ProductID, D.ProductID FROM AdventureWorks.Sales.SalesOrderDetail D WITH (TABLOCK); The query itself is a simple join of the four tables: SELECT P.ProductMainID AS PID, P.Name, D.OrderQty, H.SalesOrderNumber, H.OrderDate, C.TerritoryID FROM #Prods P JOIN #OrdDetail D ON P.ProductMainID = D.ProductMainID AND P.ProductSubID = D.ProductSubID AND P.ProductSubSubID = D.ProductSubSubID JOIN #OrdHeader H ON D.SalesOrderID = H.SalesOrderID JOIN #Custs C ON H.CustomerID = C.CustomerID ORDER BY P.ProductMainID ASC OPTION (RECOMPILE, MAXDOP 1); Remember that these tables have no indexes at all, and only the single-column sampled statistics SQL Server automatically creates (assuming default settings).  The estimated query plan produced for the test query looks like this (click to enlarge): The Problem The problem here is one of cardinality estimation – the number of rows SQL Server expects to find at each step of the plan.  The lack of indexes and useful statistical information means that SQL Server does not have the information it needs to make a good estimate.  Every join in the plan shown above estimates that it will produce just a single row as output.  Brad covers the factors that lead to the low estimates in his post. In reality, the join between the #Prods and #OrdDetail tables will produce 121,317 rows.  It should not surprise you that this has rather dire consequences for the remainder of the query plan.  In particular, it makes a nonsense of the optimizer’s decision to use Nested Loops to join to the two remaining tables.  Instead of scanning the #OrdHeader and #Custs tables once (as it expected), it has to perform 121,317 full scans of each.  The query takes somewhere in the region of twenty minutes to run to completion on my development machine. A Solution At this point, you may be thinking the same thing I was: if we really are stuck with no indexes, the best we can do is to use hash joins everywhere. We can force the exclusive use of hash joins in several ways, the two most common being join and query hints.  A join hint means writing the query using the INNER HASH JOIN syntax; using a query hint involves adding OPTION (HASH JOIN) at the bottom of the query.  The difference is that using join hints also forces the order of the join, whereas the query hint gives the optimizer freedom to reorder the joins at its discretion. Adding the OPTION (HASH JOIN) hint results in this estimated plan: That produces the correct output in around seven seconds, which is quite an improvement!  As a purely practical matter, and given the rigid rules of the environment we find ourselves in, we might leave things there.  (We can improve the hashing solution a bit – I’ll come back to that later on). Faster Nested Loops It might surprise you to hear that we can beat the performance of the hash join solution shown above using nested loops joins exclusively, and without breaking the rules we have been set. The key to this part is to realize that a condition like (A = B) can be expressed as (A <= B) AND (A >= B).  Armed with this tremendous new insight, we can rewrite the join predicates like so: SELECT P.ProductMainID AS PID, P.Name, D.OrderQty, H.SalesOrderNumber, H.OrderDate, C.TerritoryID FROM #OrdDetail D JOIN #OrdHeader H ON D.SalesOrderID >= H.SalesOrderID AND D.SalesOrderID <= H.SalesOrderID JOIN #Custs C ON H.CustomerID >= C.CustomerID AND H.CustomerID <= C.CustomerID JOIN #Prods P ON P.ProductMainID >= D.ProductMainID AND P.ProductMainID <= D.ProductMainID AND P.ProductSubID = D.ProductSubID AND P.ProductSubSubID = D.ProductSubSubID ORDER BY D.ProductMainID OPTION (RECOMPILE, LOOP JOIN, MAXDOP 1, FORCE ORDER); I’ve also added LOOP JOIN and FORCE ORDER query hints to ensure that only nested loops joins are used, and that the tables are joined in the order they appear.  The new estimated execution plan is: This new query runs in under 2 seconds. Why Is It Faster? The main reason for the improvement is the appearance of the eager Index Spools, which are also known as index-on-the-fly spools.  If you read my Inside The Optimiser series you might be interested to know that the rule responsible is called JoinToIndexOnTheFly. An eager index spool consumes all rows from the table it sits above, and builds a index suitable for the join to seek on.  Taking the index spool above the #Custs table as an example, it reads all the CustomerID and TerritoryID values with a single scan of the table, and builds an index keyed on CustomerID.  The term ‘eager’ means that the spool consumes all of its input rows when it starts up.  The index is built in a work table in tempdb, has no associated statistics, and only exists until the query finishes executing. The result is that each unindexed table is only scanned once, and just for the columns necessary to build the temporary index.  From that point on, every execution of the inner side of the join is answered by a seek on the temporary index – not the base table. A second optimization is that the sort on ProductMainID (required by the ORDER BY clause) is performed early, on just the rows coming from the #OrdDetail table.  The optimizer has a good estimate for the number of rows it needs to sort at that stage – it is just the cardinality of the table itself.  The accuracy of the estimate there is important because it helps determine the memory grant given to the sort operation.  Nested loops join preserves the order of rows on its outer input, so sorting early is safe.  (Hash joins do not preserve order in this way, of course). The extra lazy spool on the #Prods branch is a further optimization that avoids executing the seek on the temporary index if the value being joined (the ‘outer reference’) hasn’t changed from the last row received on the outer input.  It takes advantage of the fact that rows are still sorted on ProductMainID, so if duplicates exist, they will arrive at the join operator one after the other. The optimizer is quite conservative about introducing index spools into a plan, because creating and dropping a temporary index is a relatively expensive operation.  It’s presence in a plan is often an indication that a useful index is missing. I want to stress that I rewrote the query in this way primarily as an educational exercise – I can’t imagine having to do something so horrible to a production system. Improving the Hash Join I promised I would return to the solution that uses hash joins.  You might be puzzled that SQL Server can create three new indexes (and perform all those nested loops iterations) faster than it can perform three hash joins.  The answer, again, is down to the poor information available to the optimizer.  Let’s look at the hash join plan again: Two of the hash joins have single-row estimates on their build inputs.  SQL Server fixes the amount of memory available for the hash table based on this cardinality estimate, so at run time the hash join very quickly runs out of memory. This results in the join spilling hash buckets to disk, and any rows from the probe input that hash to the spilled buckets also get written to disk.  The join process then continues, and may again run out of memory.  This is a recursive process, which may eventually result in SQL Server resorting to a bailout join algorithm, which is guaranteed to complete eventually, but may be very slow.  The data sizes in the example tables are not large enough to force a hash bailout, but it does result in multiple levels of hash recursion.  You can see this for yourself by tracing the Hash Warning event using the Profiler tool. The final sort in the plan also suffers from a similar problem: it receives very little memory and has to perform multiple sort passes, saving intermediate runs to disk (the Sort Warnings Profiler event can be used to confirm this).  Notice also that because hash joins don’t preserve sort order, the sort cannot be pushed down the plan toward the #OrdDetail table, as in the nested loops plan. Ok, so now we understand the problems, what can we do to fix it?  We can address the hash spilling by forcing a different order for the joins: SELECT P.ProductMainID AS PID, P.Name, D.OrderQty, H.SalesOrderNumber, H.OrderDate, C.TerritoryID FROM #Prods P JOIN #Custs C JOIN #OrdHeader H ON H.CustomerID = C.CustomerID JOIN #OrdDetail D ON D.SalesOrderID = H.SalesOrderID ON P.ProductMainID = D.ProductMainID AND P.ProductSubID = D.ProductSubID AND P.ProductSubSubID = D.ProductSubSubID ORDER BY D.ProductMainID OPTION (MAXDOP 1, HASH JOIN, FORCE ORDER); With this plan, each of the inputs to the hash joins has a good estimate, and no hash recursion occurs.  The final sort still suffers from the one-row estimate problem, and we get a single-pass sort warning as it writes rows to disk.  Even so, the query runs to completion in three or four seconds.  That’s around half the time of the previous hashing solution, but still not as fast as the nested loops trickery. Final Thoughts SQL Server’s optimizer makes cost-based decisions, so it is vital to provide it with accurate information.  We can’t really blame the performance problems highlighted here on anything other than the decision to use completely unindexed tables, and not to allow the creation of additional statistics. I should probably stress that the nested loops solution shown above is not one I would normally contemplate in the real world.  It’s there primarily for its educational and entertainment value.  I might perhaps use it to demonstrate to the sceptical that SQL Server itself is crying out for an index. Be sure to read Brad’s original post for more details.  My grateful thanks to him for granting permission to reuse some of his material. Paul White Email: [email protected] Twitter: @PaulWhiteNZ

    Read the article

  • How to Reuse Your Old Wi-Fi Router as a Network Switch

    - by Jason Fitzpatrick
    Just because your old Wi-Fi router has been replaced by a newer model doesn’t mean it needs to gather dust in the closet. Read on as we show you how to take an old and underpowered Wi-Fi router and turn it into a respectable network switch (saving your $20 in the process). Image by mmgallan. Why Do I Want To Do This? Wi-Fi technology has changed significantly in the last ten years but Ethernet-based networking has changed very little. As such, a Wi-Fi router with 2006-era guts is lagging significantly behind current Wi-Fi router technology, but the Ethernet networking component of the device is just as useful as ever; aside from potentially being only 100Mbs instead of 1000Mbs capable (which for 99% of home applications is irrelevant) Ethernet is Ethernet. What does this matter to you, the consumer? It means that even though your old router doesn’t hack it for your Wi-Fi needs any longer the device is still a perfectly serviceable (and high quality) network switch. When do you need a network switch? Any time you want to share an Ethernet cable among multiple devices, you need a switch. For example, let’s say you have a single Ethernet wall jack behind your entertainment center. Unfortunately you have four devices that you want to link to your local network via hardline including your smart HDTV, DVR, Xbox, and a little Raspberry Pi running XBMC. Instead of spending $20-30 to purchase a brand new switch of comparable build quality to your old Wi-Fi router it makes financial sense (and is environmentally friendly) to invest five minutes of your time tweaking the settings on the old router to turn it from a Wi-Fi access point and routing tool into a network switch–perfect for dropping behind your entertainment center so that your DVR, Xbox, and media center computer can all share an Ethernet connection. What Do I Need? For this tutorial you’ll need a few things, all of which you likely have readily on hand or are free for download. To follow the basic portion of the tutorial, you’ll need the following: 1 Wi-Fi router with Ethernet ports 1 Computer with Ethernet jack 1 Ethernet cable For the advanced tutorial you’ll need all of those things, plus: 1 copy of DD-WRT firmware for your Wi-Fi router We’re conducting the experiment with a Linksys WRT54GL Wi-Fi router. The WRT54 series is one of the best selling Wi-Fi router series of all time and there’s a good chance a significant number of readers have one (or more) of them stuffed in an office closet. Even if you don’t have one of the WRT54 series routers, however, the principles we’re outlining here apply to all Wi-Fi routers; as long as your router administration panel allows the necessary changes you can follow right along with us. A quick note on the difference between the basic and advanced versions of this tutorial before we proceed. Your typical Wi-Fi router has 5 Ethernet ports on the back: 1 labeled “Internet”, “WAN”, or a variation thereof and intended to be connected to your DSL/Cable modem, and 4 labeled 1-4 intended to connect Ethernet devices like computers, printers, and game consoles directly to the Wi-Fi router. When you convert a Wi-Fi router to a switch, in most situations, you’ll lose two port as the “Internet” port cannot be used as a normal switch port and one of the switch ports becomes the input port for the Ethernet cable linking the switch to the main network. This means, referencing the diagram above, you’d lose the WAN port and LAN port 1, but retain LAN ports 2, 3, and 4 for use. If you only need to switch for 2-3 devices this may be satisfactory. However, for those of you that would prefer a more traditional switch setup where there is a dedicated WAN port and the rest of the ports are accessible, you’ll need to flash a third-party router firmware like the powerful DD-WRT onto your device. Doing so opens up the router to a greater degree of modification and allows you to assign the previously reserved WAN port to the switch, thus opening up LAN ports 1-4. Even if you don’t intend to use that extra port, DD-WRT offers you so many more options that it’s worth the extra few steps. Preparing Your Router for Life as a Switch Before we jump right in to shutting down the Wi-Fi functionality and repurposing your device as a network switch, there are a few important prep steps to attend to. First, you want to reset the router (if you just flashed a new firmware to your router, skip this step). Following the reset procedures for your particular router or go with what is known as the “Peacock Method” wherein you hold down the reset button for thirty seconds, unplug the router and wait (while still holding the reset button) for thirty seconds, and then plug it in while, again, continuing to hold down the rest button. Over the life of a router there are a variety of changes made, big and small, so it’s best to wipe them all back to the factory default before repurposing the router as a switch. Second, after resetting, we need to change the IP address of the device on the local network to an address which does not directly conflict with the new router. The typical default IP address for a home router is 192.168.1.1; if you ever need to get back into the administration panel of the router-turned-switch to check on things or make changes it will be a real hassle if the IP address of the device conflicts with the new home router. The simplest way to deal with this is to assign an address close to the actual router address but outside the range of addresses that your router will assign via the DHCP client; a good pick then is 192.168.1.2. Once the router is reset (or re-flashed) and has been assigned a new IP address, it’s time to configure it as a switch. Basic Router to Switch Configuration If you don’t want to (or need to) flash new firmware onto your device to open up that extra port, this is the section of the tutorial for you: we’ll cover how to take a stock router, our previously mentioned WRT54 series Linksys, and convert it to a switch. Hook the Wi-Fi router up to the network via one of the LAN ports (consider the WAN port as good as dead from this point forward, unless you start using the router in its traditional function again or later flash a more advanced firmware to the device, the port is officially retired at this point). Open the administration control panel via  web browser on a connected computer. Before we get started two things: first,  anything we don’t explicitly instruct you to change should be left in the default factory-reset setting as you find it, and two, change the settings in the order we list them as some settings can’t be changed after certain features are disabled. To start, let’s navigate to Setup ->Basic Setup. Here you need to change the following things: Local IP Address: [different than the primary router, e.g. 192.168.1.2] Subnet Mask: [same as the primary router, e.g. 255.255.255.0] DHCP Server: Disable Save with the “Save Settings” button and then navigate to Setup -> Advanced Routing: Operating Mode: Router This particular setting is very counterintuitive. The “Operating Mode” toggle tells the device whether or not it should enable the Network Address Translation (NAT)  feature. Because we’re turning a smart piece of networking hardware into a relatively dumb one, we don’t need this feature so we switch from Gateway mode (NAT on) to Router mode (NAT off). Our next stop is Wireless -> Basic Wireless Settings: Wireless SSID Broadcast: Disable Wireless Network Mode: Disabled After disabling the wireless we’re going to, again, do something counterintuitive. Navigate to Wireless -> Wireless Security and set the following parameters: Security Mode: WPA2 Personal WPA Algorithms: TKIP+AES WPA Shared Key: [select some random string of letters, numbers, and symbols like JF#d$di!Hdgio890] Now you may be asking yourself, why on Earth are we setting a rather secure Wi-Fi configuration on a Wi-Fi router we’re not going to use as a Wi-Fi node? On the off chance that something strange happens after, say, a power outage when your router-turned-switch cycles on and off a bunch of times and the Wi-Fi functionality is activated we don’t want to be running the Wi-Fi node wide open and granting unfettered access to your network. While the chances of this are next-to-nonexistent, it takes only a few seconds to apply the security measure so there’s little reason not to. Save your changes and navigate to Security ->Firewall. Uncheck everything but Filter Multicast Firewall Protect: Disable At this point you can save your changes again, review the changes you’ve made to ensure they all stuck, and then deploy your “new” switch wherever it is needed. Advanced Router to Switch Configuration For the advanced configuration, you’ll need a copy of DD-WRT installed on your router. Although doing so is an extra few steps, it gives you a lot more control over the process and liberates an extra port on the device. Hook the Wi-Fi router up to the network via one of the LAN ports (later you can switch the cable to the WAN port). Open the administration control panel via web browser on the connected computer. Navigate to the Setup -> Basic Setup tab to get started. In the Basic Setup tab, ensure the following settings are adjusted. The setting changes are not optional and are required to turn the Wi-Fi router into a switch. WAN Connection Type: Disabled Local IP Address: [different than the primary router, e.g. 192.168.1.2] Subnet Mask: [same as the primary router, e.g. 255.255.255.0] DHCP Server: Disable In addition to disabling the DHCP server, also uncheck all the DNSMasq boxes as the bottom of the DHCP sub-menu. If you want to activate the extra port (and why wouldn’t you), in the WAN port section: Assign WAN Port to Switch [X] At this point the router has become a switch and you have access to the WAN port so the LAN ports are all free. Since we’re already in the control panel, however, we might as well flip a few optional toggles that further lock down the switch and prevent something odd from happening. The optional settings are arranged via the menu you find them in. Remember to save your settings with the save button before moving onto a new tab. While still in the Setup -> Basic Setup menu, change the following: Gateway/Local DNS : [IP address of primary router, e.g. 192.168.1.1] NTP Client : Disable The next step is to turn off the radio completely (which not only kills the Wi-Fi but actually powers the physical radio chip off). Navigate to Wireless -> Advanced Settings -> Radio Time Restrictions: Radio Scheduling: Enable Select “Always Off” There’s no need to create a potential security problem by leaving the Wi-Fi radio on, the above toggle turns it completely off. Under Services -> Services: DNSMasq : Disable ttraff Daemon : Disable Under the Security -> Firewall tab, uncheck every box except “Filter Multicast”, as seen in the screenshot above, and then disable SPI Firewall. Once you’re done here save and move on to the Administration tab. Under Administration -> Management:  Info Site Password Protection : Enable Info Site MAC Masking : Disable CRON : Disable 802.1x : Disable Routing : Disable After this final round of tweaks, save and then apply your settings. Your router has now been, strategically, dumbed down enough to plod along as a very dependable little switch. Time to stuff it behind your desk or entertainment center and streamline your cabling.     

    Read the article

  • Queued Loadtest to remove Concurrency issues using Shared Data Service in OpenScript

    - by stefan.thieme(at)oracle.com
    Queued Processing to remove Concurrency issues in Loadtest ScriptsSome scripts act on information returned by the server, e.g. act on first item in the returned list of pending tasks/actions. This may lead to concurrency issues if the virtual users simulated in a load test scenario are not synchronized in some way.As the load test cases should be carried out in a comparable and straight forward manner simply cancel a transaction in case a collision occurs is clearly not an option. In case you increase the number of virtual users this approach would lead to a high number of requests for the early steps in your transaction (e.g. login, retrieve list of action points, assign an action point to the virtual user) but later steps would be rarely visited successfully or at all, depending on the application logic.A way to tackle this problem is to enqueue the virtual users in a Shared Data Service queue. Only the first virtual user in this queue will be allowed to carry out the critical steps (retrieve list of action points, assign an action point to the virtual user) in your transaction at any one time.Once a virtual user has passed the critical path it will dequeue himself from the head of the queue and continue with his actions. This does theoretically allow virtual users to run in parallel all steps of the transaction which are not part of the critical path.In practice it has been seen this is rarely the case, though it does not allow adding more than N users to perform a transaction without causing delays due to virtual users waiting in the queue. N being the time of the total transaction divided by the sum of the time of all critical steps in this transaction.While this problem can be circumvented by allowing multiple queues to act on individual segments of the list of actions, e.g. per country filter, ends with 0..9 filter, etc.This would require additional handling of these additional queues of slots for the virtual users at the head of the queue in order to maintain the mutually exclusive access to the first element in the list returned by the server at any one time of the load test. Such an improved handling of multiple queues and/or multiple slots is above the subject of this paper.Shared Data Services Pre-RequisitesStart WebLogic Server to host Shared Data ServicesYou will have to make sure that your WebLogic server is installed and started. Shared Data Services may not work if you installed only the minimal installation package for OpenScript. If however you installed the default package including OLT and OTM, you may follow the instructions below to start and verify WebLogic installation.To start the WebLogic Server deployed underneath of Oracle Load Testing and/or Oracle Test Manager you can go to your Start menu, Oracle Application Testing Suite and select the Restart Oracle Application Testing Suite Application Service entry from the Tools submenu.To verify the service has been started you can run the Microsoft Management Console for Services by Selecting Run from the Start Menu and entering services.msc. Look for the entry that reads Oracle Application Testing Suite Application Service, once it has changed it status from Starting to Started you can proceed to verify the login. Please note that this may take several minutes, I would say up to 10 minutes depending on the strength of your CPU horse-power.Verify WebLogic Server user credentialsYou will have to make sure that your WebLogic Server is installed and started. Next open the Oracle WebLogic Server Adminstration Console on http://localhost:8088/console.It may take a while until the application is deployed and started. It may display the following until the Administration Console has been deployed on the fly.Afterwards you can login using the username oats and the password that you selected during install time for your Application Testing Suite administrative purposes.This will bring up the Home page of you WebLogic Server. You have actually verified that you are able to login with these credentials already. However if you want to check the details, navigate to Security Realms, myrealm, Users and Groups tab.Here you could add users to your WebLogic Server which could be used in the later steps. Details on the Groups required for such a custom user to work are exceeding this quick overview and have to be selected with the WebLogic Server Adminstration Guide in mind.Shared Data Services pre-requisites for Load testingOpenScript Preferences have to be set to enable Encryption and provide a default Shared Data Service Connection for Playback.These are pre-requisites you want to use for load testing with Shared Data Services.Please note that the usage of the Connection Parameters (individual directive in the script) for Shared Data Services did not playback reliably in the current version 9.20.0370 of Oracle Load Testing (OLT) and encryption of credentials still seemed to be mandatory as well.General Encryption settingsSelect OpenScript Preferences from the View menu and navigate to the General, Encryption entry in the tree on the left. Select the Encrypt script data option from the list and enter the same password that you used for securing your WebLogic Server Administration Console.Enable global shared data access credentialsSelect OpenScript Preferences from the View menu and navigate to the Playback, Shared Data entry in the tree on the left. Enable the global shared data access credentials and enter the Address, User name and Password determined for your WebLogic Server to host Shared Data Services.Please note, that you may want to replace the localhost in Address with the hosts realname in case you plan to run load tests with Loadtest Agents running on remote systems.Queued Processing of TransactionsEnable Shared Data Services Module in Script PropertiesThe Shared Data Services Module has to be enabled for each Script that wants to employ the Shared Data Service Queue functionality in OpenScript. It can be enabled under the Script menu selecting Script Properties. On the Script Properties Dialog select the Modules section and check Shared Data to enable Shared Data Service Module for your script. Checking the Shared Data Services option will effectively add a line to your script code that adds the sharedData ScriptService to your script class of IteratingVUserScript.@ScriptService oracle.oats.scripting.modules.sharedData.api.SharedDataService sharedData;Record your scriptRecord your script as usual and then add the following things for Queue handling in the Initialize code block, before the first step and after the last step of your critical path and in the Finalize code block.The java code to be added at individual locations is explained in the following sections in full detail.Create a Shared Data Queue in InitializeTo create a Shared Data Queue go to the Java view of your script and enter the following statements to the initialize() code block.info("Create queueA with life time of 120 minutes");sharedData.createQueue("queueA", 120);This will create an instantiation of the Shared Data Queue object named queueA which is maintained for upto 120 minutes.If you want to use the code for multiple scripts, make sure to use a different queue name for each one here and in the subsequent steps. You may even consider to use a dynamic queueName based on filters of your result list being concurrently accessed.Prepare a unique id for each IterationIn order to keep track of individual virtual users in our queue we need to create a unique identifier from the virtual user id and the used username right after retrieving the next record from our databank file.getDatabank("Usernames").getNextDatabankRecord();getVariables().set("usernameValue1","VU_{{@vuid}}_{{@iterationnum}}_{{db.Usernames.Username}}_{{@timestamp}}_{{@random(10000)}}");String usernameValue = getVariables().get("usernameValue1");info("Now running virtual user " + usernameValue);As you can see from the above code block, we have set the OpenScript variable usernameValue1 to VU_{{@vuid}}_{{@iterationnum}}_{{db.Usernames.Username}}_{{@timestamp}}_{{@random(10000)}} which is a concatenation of the virtual user id and the iterationnumber for general uniqueness; as well as the username from our databank, the timestamp and a random number for making it further unique and ease spotting of errors.Not all of these fields are actually required to make it really unique, but adding the queue name may also be considered to help troubleshoot multiple queues.The value is then retrieved with the getVariables.get() method call and assigned to the usernameValue String used throughout the script.Please note that moving the getDatabank("Usernames").getNextDatabankRecord(); call to the initialize block was later considered to remove concurrency of multiple virtual users running with the same userid and therefor accessing the same "My Inbox" in step 6. This will effectively give each virtual user a userid from the databank file. Make sure you have enough userids to remove this second hurdle.Enqueue and attend Queue before Critical PathTo maintain the right order of virtual users being allowed into the critical path of the transaction the following pseudo step has to be added in front of the first critical step. In the case of this example this is right in front of the step where we retrieve the list of actions from which we select the first to be assigned to us.beginStep("[0] Waiting in the Queue", 0);{info("Enqueued virtual user " + usernameValue + " at the end of queueA");sharedData.offerLast("queueA", usernameValue);info("Wait until the user is the first in queueA");String queueValue1 = null;do {// we wait for at least 0.7 seconds before we check the head of the// queue. This is the time it takes one user to move through the// critical path, i.e. pass steps [5] Enter country and [6] Assign// to meThread.sleep(700);queueValue1 = (String) sharedData.peekFirst("queueA");info("The first user in queueA is currently: '" + queueValue1 + "' " + queueValue1.getClass() + " length " + queueValue1.length() );info("The current user is '"+ usernameValue + "' " + usernameValue.getClass() + " length " + usernameValue.length() + ": indexOf " + usernameValue.indexOf(queueValue1) + " equals " + usernameValue.equals(queueValue1) );} while ( queueValue1.indexOf(usernameValue) < 0 );info("Now the user is the first in queueA");}endStep();This will enqueue the username to the tail of our Queue. It will will wait for at least 700 milliseconds, the time it takes for one user to exit the critical path and then compare the head of our queue with it's username. This last step will be repeated while the two are not equal (indexOf less than zero). If they are equal the indexOf will yield a value of zero or larger and we will perform the critical steps.Dequeue after Critical PathAfter the virtual user has left the critical path and complete its last step the following code block needs to dequeue the virtual user. In the case of our example this is right after the action has been actually assigned to the virtual user. This will allow the next virtual user to retrieve the list of actions still available and in turn let him make his selection/assignment.info("Get and remove the current user from the head of queueA");String pollValue1 = (String) sharedData.pollFirst("queueA");The current user is removed from the head of the queue. The next one will now be able to match his username against the head of the queue.Clear and Destroy Queue for FinishWhen the script has completed, it should clear and destroy the queue. This code block can be put in the finish block of your script and/or in a separate script in order to clear and remove the queue in case you have spotted an error or want to reset the queue for some reason.info("Clear queueA");sharedData.clearQueue("queueA");info("Destroy queueA");sharedData.destroyQueue("queueA");The users waiting in queueA are cleared and the queue is destroyed. If you have scripts still executing they will be caught in a loop.I found it better to maintain a separate Reset Queue script which contained only the following code in the initialize() block. I use to call this script to make sure the queue is cleared in between multiple Loadtest runs. This script could also even be added as the first in a larger scenario, which would execute it only once at very start of the Loadtest and make sure the queues do not contain any stale entries.info("Create queueA with life time of 120 minutes");sharedData.createQueue("queueA", 120);info("Clear queueA");sharedData.clearQueue("queueA");This will create a Shared Data Queue instance of queueA and clear all entries from this queue.Monitoring QueueWhile creating the scripts it was useful to monitor the contents, i.e. the current first user in the Queue. The following code block will make sure the Shared Data Queue is accessible in the initialize() block.info("Create queueA with life time of 120 minutes");sharedData.createQueue("queueA", 120);In the run() block the following code will continuously monitor the first element of the Queue and write an informational message with the current username Value to the Result window.info("Monitor the first users in queueA");String queueValue1 = null;do {queueValue1 = (String) sharedData.peekFirst("queueA");if (queueValue1 != null)info("The first user in queueA is currently: '" + queueValue1 + "' " + queueValue1.getClass() + " length " + queueValue1.length() );} while ( true );This script can be run from OpenScript parallel to a loadtest performed by the Oracle Load Test.However it is not recommend to run this in a production loadtest as the performance impact is unknown. Accessing the Queue's head with the peekFirst() method has been reported with about 2 seconds response time by both OpenScript and OTL. It is advised to log a Service Request to see if this could be lowered in future releases of Application Testing Suite, as the pollFirst() and even offerLast() writing to the tail of the Queue usually returned after an average 0.1 seconds.Debugging QueueWhile debugging the scripts the following was useful to remove single entries from its head, i.e. the current first user in the Queue. The following code block will make sure the Shared Data Queue is accessible in the initialize() block.info("Create queueA with life time of 120 minutes");sharedData.createQueue("queueA", 120);In the run() block the following code will remove the first element of the Queue and write an informational message with the current username Value to the Result window.info("Get and remove the current user from the head of queueA");String pollValue1 = (String) sharedData.pollFirst("queueA");info("The first user in queueA was currently: '" + pollValue1 + "' " + pollValue1.getClass() + " length " + pollValue1.length() );ReferencesOracle Functional Testing OpenScript User's Guide Version 9.20 [E15488-05]Chapter 17 Using the Shared Data Modulehttp://download.oracle.com/otn/nt/apptesting/oats-docs-9.21.0030.zipOracle Fusion Middleware Oracle WebLogic Server Administration Console Online Help 11g Release 1 (10.3.4) [E13952-04]Administration Console Online Help - Manage users and groupshttp://download.oracle.com/docs/cd/E17904_01/apirefs.1111/e13952/taskhelp/security/ManageUsersAndGroups.htm

    Read the article

  • In Which We Demystify A Few Docupresentment Settings And Learn the Ethos of the Author

    - by Andy Little
    It's no secret that Docupresentment (part of the Oracle Documaker suite) is powerful tool for integrating on-demand and interactive applications for publishing with the Oracle Documaker framework.  It's also no secret there are are many details with respect to the configuration of Docupresentment that can elude even the most erudite of of techies.  To be sure, Docupresentment will work for you right out of the box, and in most cases will suit your needs without toying with a configuration file.  But, where's the adventure in that?   With this inaugural post to That's The Way, I'm going to introduce myself, and what my aim is with this blog.  If you didn't figure it out already by checking out my profile, my name is Andy and I've been with Oracle (nee Skywire Software nee Docucorp nee Formmaker) since the formative years of 1998.  Strangely, it doesn't seem that long ago, but it's certainly a lifetime in the age of technology.  I recall running a BBS from my parent's basement on a 1200 baud modem, and the trepidation and sweaty-palmed excitement of upgrading to the power and speed of 2400 baud!  Fine, I'll admit that perhaps I'm inflating the experience a bit, but I was kid!  This is the stuff of War Games and King's Quest I and the demise of TI-99 4/A.  Exciting times.  So fast-forward a bit and I'm 12 years into a career in the world of document automation and publishing working for the best (IMHO) software company on the planet.  With That's The Way I hope to shed a little light and peek under the covers of some of the more interesting aspects of implementations involving the tech space within the Oracle Insurance Global Business Unit (IGBU), which includes Oracle Documaker, Rating & Underwriting, and Policy Administration to name a few.  I may delve off course a bit, and you'll likely get a dose of humor (at least in my mind) but I hope you'll glean at least a tidbit of usefulness with each post.  Feel free to comment as I'm a fairly conversant guy and happy to talk -- it's stopping the talking that's the hard part... So, back to our regularly-scheduled post, already in progress.  By this time you've visited Oracle's E-Delivery site and acquired your properly-licensed version of Oracle Documaker.  Wait -- you didn't find it?  Understandable -- navigating the voluminous download library within Oracle can be a daunting task.  It's pretty simple once you’ve done it a few times.  Login to the e-delivery site, and accept the license terms and restrictions.  Then, you’ll be able to select the Oracle Insurance Applications product pack and your appropriate platform. Click Go and you’ll see a list of applicable products, and you’ll click on Oracle Documaker Media Pack (as I went to press with this article the version is 11.4): Finally, click the Download button next to Docupresentment (again, version at press time is 2.2 p5). This should give you a ZIP file that contains the installation packages for the Docupresentment Server and Client, cryptically named IDSServer22P05W32.exe and IDSClient22P05W32.exe. At this time, I’d like to take a little detour and explain that the world of Oracle, like most technical companies, is rife with acronyms.  One of the reasons Skywire Software was a appealing to Oracle was our use of many acronyms, including the occasional use of multiple acronyms with the same meaning.  I apologize in advance and will try to point these out along the way.  Here’s your first sticky note to go along with that: IDS = Internet Document Server = Docupresentment Once you’ve completed the installation, you’ll have a shiny new Docupresentment server and client, and if you installed the default location it will be living in c:\docserv. Unix users, I’m one of you!  You’ll find it by default in  ~/docupresentment/docserv.  Forging onward with the meat of this post is learning about some special configuration options.  By now you’ve read the documentation included with the download (specifically ids_book.pdf) which goes into some detail of the rubric of the configuration file and in fact there’s even a handy utility that provides an interface to the configuration file (see Running IDSConfig in the documentation).  But who wants to deal with a configuration utility when we have the tools and technology to edit the file <gasp> by hand! I shall now proceed with the standard Information Technology Under the Hood Disclaimer: Please remember to back up any files before you make changes.  I am not responsible for any havoc you may wreak! Go to your installation directory, and locate your docserv.xml file.  Open it in your favorite XML editor.  I happen to be fond of Notepad++ with the XML Tools plugin.  Almost immediately you will behold the splendor of the configuration file.  Just take a moment and let that sink in.  Ok – moving on.  If you reviewed the documentation you know that inside the root <configuration> node there are multiple <section> nodes, each containing a specific group of settings.  Let’s take a look at <section name=”DocumentServer”>: There are a few entries I’d like to discuss.  First, <entry name=”StartCommand”>. This should be pretty self-explanatory; it’s the name of the executable that’s run when you fire up Docupresentment.  Immediately following that is <entry name=”StartArguments”> and as you might imagine these are the arguments passed to the executable.  A few things to point out: The –Dids.configuration=docserv.xml parameter specifies the name of your configuration file. The –Dlogging.configuration=logconf.xml parameter specifies the name of your logging configuration file (this uses log4j so bone up on that before you delve here). The -Djava.endorsed.dirs=lib/endorsed parameter specifies the path where 3rd party Java libraries can be located for use with Docupresentment.  More on that in another post. The <entry name=”Instances”> allows you to specify the number of instances of Docupresentment that will be started.  By default this is two, and generally two instances per CPU is adequate, however you will always need to perform load testing to determine the sweet spot based on your hardware and types of transactions.  You may have many, many more instances than 2. Time for a sidebar on instances.  An instance is nothing more than a separate process of Docupresentment.  The Docupresentment service that you fire up with docserver.bat or docserver.sh actually starts a watchdog process, which is then responsible for starting up the actual Docupresentment processes.  Each of these act independently from one another, so if one crashes, it does not affect any others.  In the case of a crashed process, the watchdog will start up another instance so the number of configured instances are always running.  Bottom line: instance = Docupresentment process. And now, finally, to the settings which gave me pause on an not-too-long-ago implementation!  Docupresentment includes a feature that watches configuration files (such as docserv.xml and logconf.xml) and will automatically restart its instances to load the changes.  You can configure the time that Docupresentment waits to check these files using the setting <entry name=”FileWatchTimeMillis”>.  By default the number is 12000ms, or 12 seconds.  You can save yourself a few CPU cycles by extending this time, or by disabling  the check altogether by setting the value to 0.  This may or may not be appropriate for your environment; if you have 100% uptime requirements then you probably don’t want to bring down an entire set of processes just to accept a new configuration value, so it’s best to leave this somewhere between 12 seconds to a few minutes.  Another point to keep in mind: if you are using Documaker real-time processing under Docupresentment the Master Resource Library (MRL) files and INI options are cached, and if you need to affect a change, you’ll have to “restart” Docupresentment.  Touching the docserv.xml file is an easy way to do this (other methods including using the RSS request, but that’s another post). The next item up: <entry name=”FilePurgeTimeSeconds”>.  You may already know that the Docupresentment system can generate many temporary files based on certain request types that are processed through the system.  What you may not know is how those files are cleaned up.  There are many rules in Docupresentment that cause the creation of temporary files.  When these files are created, Docupresentment writes an entry into a properties file called the file cache.  This file contains the name, creation date, and expiration time of each temporary file created by each instance of Docupresentment.  Periodically Docupresentment will check the file cache to determine if there are files that are past the expiration time, not unlike that block of cheese festering away in the back of my refrigerator.  However, unlike my ‘fridge cleaning tendencies, Docupresentment is quick to remove files that are past their expiration time.  You, my friend, have the power to control how often Docupresentment inspects the file cache.  Simply set the value for <entry name=”FilePurgeTimeSeconds”> to the number of seconds appropriate for your requirements and you’re set.  Note that file purging happens on a separate thread from normal request processing, so this shouldn’t interfere with response times unless the CPU happens to be really taxed at the point of cache processing.  Finally, after all of this, we get to the final setting I’m going to address in this post: <entry name=”FilePurgeList”>.  The default is “filecache.properties”.  This establishes the root name for the Docupresentment file cache that I mentioned previously.  Docupresentment creates a separate cache file for each instance based on this setting.  If you have two instances, you’ll see two files created: filecache.properties.1 and filecache.properties.2.  Feel free to open these up and check them out. I hope you’ve enjoyed this first foray into the configuration file of Docupresentment.  If you did enjoy it, feel free to drop a comment, I welcome feedback.  If you have ideas for other posts you’d like to see, please do let me know.  You can reach me at [email protected]. ‘Til next time! ###

    Read the article

  • New Enhancements for InnoDB Memcached

    - by Calvin Sun
    In MySQL 5.6, we continued our development on InnoDB Memcached and completed a few widely desirable features that make InnoDB Memcached a competitive feature in more scenario. Notablely, they are 1) Support multiple table mapping 2) Added background thread to auto-commit long running transactions 3) Enhancement in binlog performance  Let’s go over each of these features one by one. And in the last section, we will go over a couple of internally performed performance tests. Support multiple table mapping In our earlier release, all InnoDB Memcached operations are mapped to a single InnoDB table. In the real life, user might want to use this InnoDB Memcached features on different tables. Thus being able to support access to different table at run time, and having different mapping for different connections becomes a very desirable feature. And in this GA release, we allow user just be able to do both. We will discuss the key concepts and key steps in using this feature. 1) "mapping name" in the "get" and "set" command In order to allow InnoDB Memcached map to a new table, the user (DBA) would still require to "pre-register" table(s) in InnoDB Memcached “containers” table (there is security consideration for this requirement). If you would like to know about “containers” table, please refer to my earlier blogs in blogs.innodb.com. Once registered, the InnoDB Memcached will then be able to look for such table when they are referred. Each of such registered table will have a unique "registration name" (or mapping_name) corresponding to the “name” field in the “containers” table.. To access these tables, user will include such "registration name" in their get or set commands, in the form of "get @@new_mapping_name.key", prefix "@@" is required for signaling a mapped table change. The key and the "mapping name" are separated by a configurable delimiter, by default, it is ".". So the syntax is: get [@@mapping_name.]key_name set [@@mapping_name.]key_name  or  get @@mapping_name set @@mapping_name Here is an example: Let's set up three tables in the "containers" table: The first is a map to InnoDB table "test/demo_test" table with mapping name "setup_1" INSERT INTO containers VALUES ("setup_1", "test", "demo_test", "c1", "c2", "c3", "c4", "c5", "PRIMARY");  Similarly, we set up table mappings for table "test/new_demo" with name "setup_2" and that to table "mydatabase/my_demo" with name "setup_3": INSERT INTO containers VALUES ("setup_2", "test", "new_demo", "c1", "c2", "c3", "c4", "c5", "secondary_index_x"); INSERT INTO containers VALUES ("setup_3", "my_database", "my_demo", "c1", "c2", "c3", "c4", "c5", "idx"); To switch to table "my_database/my_demo", and get the value corresponding to “key_a”, user will do: get @@setup_3.key_a (this will also output the value that corresponding to key "key_a" or simply get @@setup_3 Once this is done, this connection will switch to "my_database/my_demo" table until another table mapping switch is requested. so it can continue issue regular command like: get key_b  set key_c 0 0 7 These DMLs will all be directed to "my_database/my_demo" table. And this also implies that different connections can have different bindings (to different table). 2) Delimiter: For the delimiter "." that separates the "mapping name" and key value, we also added a configure option in the "config_options" system table with name of "table_map_delimiter": INSERT INTO config_options VALUES("table_map_delimiter", "."); So if user wants to change to a different delimiter, they can change it in the config_option table. 3) Default mapping: Once we have multiple table mapping, there should be always a "default" map setting. For this, we decided if there exists a mapping name of "default", then this will be chosen as default mapping. Otherwise, the first row of the containers table will chosen as default setting. Please note, user tables can be repeated in the "containers" table (for example, user wants to access different columns of the table in different settings), as long as they are using different mapping/configure names in the first column, which is enforced by a unique index. 4) bind command In addition, we also extend the protocol and added a bind command, its usage is fairly straightforward. To switch to "setup_3" mapping above, you simply issue: bind setup_3 This will switch this connection's InnoDB table to "my_database/my_demo" In summary, with this feature, you now can direct access to difference tables with difference session. And even a single connection, you can query into difference tables. Background thread to auto-commit long running transactions This is a feature related to the “batch” concept we discussed in earlier blogs. This “batch” feature allows us batch the read and write operations, and commit them only after certain calls. The “batch” size is controlled by the configure parameter “daemon_memcached_w_batch_size” and “daemon_memcached_r_batch_size”. This could significantly boost performance. However, it also comes with some disadvantages, for example, you will not be able to view “uncommitted” operations from SQL end unless you set transaction isolation level to read_uncommitted, and in addition, this will held certain row locks for extend period of time that might reduce the concurrency. To deal with this, we introduce a background thread that “auto-commits” the transaction if they are idle for certain amount of time (default is 5 seconds). The background thread will wake up every second and loop through every “connections” opened by Memcached, and check for idle transactions. And if such transaction is idle longer than certain limit and not being used, it will commit such transactions. This limit is configurable by change “innodb_api_bk_commit_interval”. Its default value is 5 seconds, and minimum is 1 second, and maximum is 1073741824 seconds. With the help of such background thread, you will not need to worry about long running uncommitted transactions when set daemon_memcached_w_batch_size and daemon_memcached_r_batch_size to a large number. This also reduces the number of locks that could be held due to long running transactions, and thus further increase the concurrency. Enhancement in binlog performance As you might all know, binlog operation is not done by InnoDB storage engine, rather it is handled in the MySQL layer. In order to support binlog operation through InnoDB Memcached, we would have to artificially create some MySQL constructs in order to access binlog handler APIs. In previous lab release, for simplicity consideration, we open and destroy these MySQL constructs (such as THD) for each operations. This required us to set the “batch” size always to 1 when binlog is on, no matter what “daemon_memcached_w_batch_size” and “daemon_memcached_r_batch_size” are configured to. This put a big restriction on our capability to scale, and also there are quite a bit overhead in creating destroying such constructs that bogs the performance down. With this release, we made necessary change that would keep MySQL constructs as long as they are valid for a particular connection. So there will not be repeated and redundant open and close (table) calls. And now even with binlog option is enabled (with innodb_api_enable_binlog,), we still can batch the transactions with daemon_memcached_w_batch_size and daemon_memcached_r_batch_size, thus scale the write/read performance. Although there are still overheads that makes InnoDB Memcached cannot perform as fast as when binlog is turned off. It is much better off comparing to previous release. And we are continuing optimize the solution is this area to improve the performance as much as possible. Performance Study: Amerandra of our System QA team have conducted some performance studies on queries through our InnoDB Memcached connection and plain SQL end. And it shows some interesting results. The test is conducted on a “Linux 2.6.32-300.7.1.el6uek.x86_64 ix86 (64)” machine with 16 GB Memory, Intel Xeon 2.0 GHz CPU X86_64 2 CPUs- 4 Core Each, 2 RAID DISKS (1027 GB,733.9GB). Results are described in following tables: Table 1: Performance comparison on Set operations Connections 5.6.7-RC-Memcached-plugin ( TPS / Qps) with memcached-threads=8*** 5.6.7-RC* X faster Set (QPS) Set** 8 30,000 5,600 5.36 32 59,000 13,000 4.54 128 68,000 8,000 8.50 512 63,000 6.800 9.23 * mysql-5.6.7-rc-linux2.6-x86_64 ** The “set” operation when implemented in InnoDB Memcached involves a couple of DMLs: it first query the table to see whether the “key” exists, if it does not, the new key/value pair will be inserted. If it does exist, the “value” field of matching row (by key) will be updated. So when used in above query, it is a precompiled store procedure, and query will just execute such procedures. *** added “–daemon_memcached_option=-t8” (default is 4 threads) So we can see with this “set” query, InnoDB Memcached can run 4.5 to 9 time faster than MySQL server. Table 2: Performance comparison on Get operations Connections 5.6.7-RC-Memcached-plugin ( TPS / Qps) with memcached-threads=8 5.6.7-RC* X faster Get (QPS) Get 8 42,000 27,000 1.56 32 101,000 55.000 1.83 128 117,000 52,000 2.25 512 109,000 52,000 2.10 With the “get” query (or the select query), memcached performs 1.5 to 2 times faster than normal SQL. Summary: In summary, we added several much-desired features to InnoDB Memcached in this release, allowing user to operate on different tables with this Memcached interface. We also now provide a background commit thread to commit long running idle transactions, thus allow user to configure large batch write/read without worrying about large number of rows held or not being able to see (uncommit) data. We also greatly enhanced the performance when Binlog is enabled. We will continue making efforts in both performance enhancement and functionality areas to make InnoDB Memcached a good demo case for our InnoDB APIs. Jimmy Yang, September 29, 2012

    Read the article

  • When was sys.dm_os_wait_stats last cleared?

    - by SQLOS Team
    The sys.dm_os_wait_stats DMV provides essential metrics for diagnosing SQL Server performance problems. Returning incrementally accumulating information about all the completed waits encountered by executing threads it is a useful way to identify bottlenecks such as IO latency issues or waits on locks. The counters are reset each time SQL server is restarted, or when the following command is run: DBCC SQLPERF ('sys.dm_os_wait_stats', CLEAR); To make sense out of these wait values you need to know how they change over time. Suppose you are asked to troubleshoot a system and you don't know when the wait stats were last zeroed. Is there any way to find the elapsed time since this happened? If the wait stats were not cleared using the DBCC SQLPERF command then you can simply correlate the stats with the time SQL Server was started using the sqlserver_start_time column introduced in SQL Server 2008 R2: SELECT sqlserver_start_time from sys.dm_os_sys_info However how do you tell if someone has run DBCC SQLPERF ('sys.dm_os_wait_stats', CLEAR) since the server was started, and if they did, when? Without this information the initial, or historical, wait_stats have less value until you can measure deltas over time. There is a way to at least estimate when the stats were last cleared, by using the wait stats themselves and choosing a thread that spends most of its time sleeping. A good candidate is the SQL Trace incremental flush task, which mostly sleeps (in 4 second intervals) and in between it attempts to flush (if there are new events – which is rare when only default trace is running) – so it pretty much sleeps all the time. Hence the time it has spent waiting is very close to the elapsed time since the counter was reset. Credit goes to Ivan Penkov in the SQLOS dev team for suggesting this. Here's an example (excuse formatting): 144 seconds after the server was started: select top 10 wait_type, wait_time_ms from sys.dm_os_wait_stats order by wait_time_ms desc wait_type                                                               wait_time_ms--------------------------------------------------------------------------------------------------------------- XE_DISPATCHER_WAIT                                      242273LAZYWRITER_SLEEP                                          146010LOGMGR_QUEUE                                                145412DIRTY_PAGE_POLL                                             145411XE_TIMER_EVENT                                               145216REQUEST_FOR_DEADLOCK_SEARCH             145194SQLTRACE_INCREMENTAL_FLUSH_SLEEP    144325SLEEP_TASK                                                        73359BROKER_TO_FLUSH                                           73113PREEMPTIVE_OS_AUTHENTICATIONOPS       143 (10 rows affected) Reset: DBCC SQLPERF('sys.dm_os_wait_stats', CLEAR)" DBCC execution completed. If DBCC printed error messages, contact your system administrator. After 8 seconds: select top 10 wait_type, wait_time_ms from sys.dm_os_wait_stats order by wait_time_ms desc wait_type                                                                 wait_time_ms--------------------------------------------------------------------------------------------------------------------- REQUEST_FOR_DEADLOCK_SEARCH              10013LAZYWRITER_SLEEP                                           8124SQLTRACE_INCREMENTAL_FLUSH_SLEEP     8017LOGMGR_QUEUE                                                 7579DIRTY_PAGE_POLL                                              7532XE_TIMER_EVENT                                                5007BROKER_TO_FLUSH                                            4118SLEEP_TASK                                                         3089PREEMPTIVE_OS_AUTHENTICATIONOPS        28SOS_SCHEDULER_YIELD                                   27 (10 rows affected)   After 12 seconds: select top 10 wait_type, wait_time_ms from sys.dm_os_wait_stats order by wait_time_ms desc wait_type                                                                  wait_time_ms------------------------------------------------------------------------------------------------------ REQUEST_FOR_DEADLOCK_SEARCH               15020LAZYWRITER_SLEEP                                            14206LOGMGR_QUEUE                                                  14036DIRTY_PAGE_POLL                                               13973SQLTRACE_INCREMENTAL_FLUSH_SLEEP      12026XE_TIMER_EVENT                                                 10014SLEEP_TASK                                                          7207BROKER_TO_FLUSH                                             7207PREEMPTIVE_OS_AUTHENTICATIONOPS         57SOS_SCHEDULER_YIELD                                     28 (10 rows affected) It may not be accurate to the millisecond, but it can provide a useful data point, and give an indication whether the wait stats were manually cleared after startup, and if so approximately when. - Guy     Originally posted at http://blogs.msdn.com/b/sqlosteam/

    Read the article

  • SQL Server Configuration timeouts - and a workaround [SSIS]

    - by jamiet
    Ever since I started writing SSIS packages back in 2004 I have opted to store configurations in .dtsConfig (.i.e. XML) files rather than in a SQL Server table (aka SQL Server Configurations) however recently I inherited some packages that used SQL Server Configurations and thus had to immerse myself in their murky little world. To all the people that have ever gone onto the SSIS forum and asked questions about ambiguous behaviour of SQL Server Configurations I now say this... I feel your pain! The biggest problem I have had was in dealing with the change to the order in which configurations get applied that came about in SSIS 2008. Those changes are detailed on MSDN at SSIS Package Configurations however the pertinent bits are: As the utility loads and runs the package, events occur in the following order: The dtexec utility loads the package. The utility applies the configurations that were specified in the package at design time and in the order that is specified in the package. (The one exception to this is the Parent Package Variables configurations. The utility applies these configurations only once and later in the process.) The utility then applies any options that you specified on the command line. The utility then reloads the configurations that were specified in the package at design time and in the order specified in the package. (Again, the exception to this rule is the Parent Package Variables configurations). The utility uses any command-line options that were specified to reload the configurations. Therefore, different values might be reloaded from a different location. The utility applies the Parent Package Variable configurations. The utility runs the package. To understand how these steps differ from SSIS 2005 I recommend reading Doug Laudenschlager’s blog post Understand how SSIS package configurations are applied. The very nature of SQL Server Configurations means that the Connection String for the database holding the configuration values needs to be supplied from the command-line. Typically then the call to execute your package resembles this: dtexec /FILE Package.dtsx /SET "\Package.Connections[SSISConfigurations].Properties[ConnectionString]";"\"Data Source=SomeServer;Initial Catalog=SomeDB;Integrated Security=SSPI;\"", The problem then is that, as per the steps above, the package will (1) attempt to apply all configurations using the Connection String stored in the package for the "SSISConfigurations" Connection Manager before then (2) applying the Connection String from the command-line and then (3) apply the same configurations all over again. In the packages that I inherited that first attempt to apply the configurations would timeout (not unexpected); I had 8 SQL Server Configurations in the package and thus the package was waiting for 2 minutes until all the Configurations timed out (i.e. 15seconds per Configuration) - in a package that only executes for ~8seconds when it gets to do its actual work a delay of 2minutes was simply unacceptable. We had three options in how to deal with this: Get rid of the use of SQL Server configurations and use .dtsConfig files instead Edit the packages when they get deployed Change the timeout on the "SSISConfigurations" Connection Manager #1 was my preferred choice but, for reasons I explain below*, wasn't an option in this particular instance. #2 was discounted out of hand because it negates the point of using Configurations in the first place. This left us with #3 - change the timeout on the Connection Manager. This is done by going into the properties of the Connection Manager, opening the "All" tab and changing the Connect Timeout property to some suitable value (in the screenshot below I chose 2 seconds). This change meant that the attempts to apply the SQL Server configurations timed out in 16 seconds rather than two minutes; clearly this isn't an optimum solution but its certainly better than it was. So there you have it - if you are having problems with SQL Server configuration timeouts within SSIS try changing the timeout of the Connection Manager. Better still - don't bother using SQL Server Configuration in the first place. Even better - install RC0 of SQL Server 2012 to start leveraging SSIS parameters and leave the nasty old world of configurations behind you. @Jamiet * Basically, we are leveraging a SSIS execution/logging framework in which the client had invested a lot of resources and SQL Server Configurations are an integral part of that.

    Read the article

  • Automating XNA Performance Testing?

    - by Grofit
    I was wondering what peoples approaches or thoughts were on automating performance testing in XNA. Currently I am looking at only working in 2d, but that poses many areas where performance can be improved with different implementations. An example would be if you had 2 different implementations of spatial partitioning, one may be faster than another but without doing some actual performance testing you wouldn't be able to tell which one for sure (unless you saw the code was blatantly slow in certain parts). You could write a unit test which for a given time frame kept adding/updating/removing entities for both implementations and see how many were made in each timeframe and the higher one would be the faster one (in this given example). Another higher level example would be if you wanted to see how many entities you can have on the screen roughly without going beneath 60fps. The problem with this is to automate it you would need to use the hidden form trick or some other thing to kick off a mock game and purely test which parts you care about and disable everything else. I know that this isnt a simple affair really as even if you can automate the tests, really it is up to a human to interpret if the results are performant enough, but as part of a build step you could have it run these tests and publish the results somewhere for comparison. This way if you go from version 1.1 to 1.2 but have changed a few underlying algorithms you may notice that generally the performance score would have gone up, meaning you have improved your overall performance of the application, and then from 1.2 to 1.3 you may notice that you have then dropped overall performance a bit. So has anyone automated this sort of thing in their projects, and if so how do you measure your performance comparisons at a high level and what frameworks do you use to test? As providing you have written your code so its testable/mockable for most parts you can just use your tests as a mechanism for getting some performance results... === Edit === Just for clarity, I am more interested in the best way to make use of automated tests within XNA to track your performance, not play testing or guessing by manually running your game on a machine. This is completely different to seeing if your game is playable on X hardware, it is more about tracking the change in performance as your game engine/framework changes. As mentioned in one of the comments you could easily test "how many nodes can I insert/remove/update within QuadTreeA within 2 seconds", but you have to physically look at these results every time to see if it has changed, which may be fine and is still better than just relying on playing it to see if you notice any difference between version. However if you were to put an Assert in to notify you of a fail if it goes lower than lets say 5000 in 2 seconds you have a brittle test as it is then contextual to the hardware, not just the implementation. Although that being said these sort of automated tests are only really any use if you are running your tests as some sort of build pipeline i.e: Checkout - Run Unit Tests - Run Integration Tests - Run Performance Tests - Package So then you can easily compare the stats from one build to another on the CI server as a report of some sort, and again this may not mean much to anyone if you are not used to Continuous Integration. The main crux of this question is to see how people manage this between builds and how they find it best to report upon. As I said it can be subjective but as knowledge will be gained from the answers it seems a worthwhile question.

    Read the article

  • Inequality joins, Asynchronous transformations and Lookups : SSIS

    - by jamiet
    It is pretty much accepted by SQL Server Integration Services (SSIS) developers that synchronous transformations are generally quicker than asynchronous transformations (for a description of synchronous and asynchronous transformations go read Asynchronous and synchronous data flow components). Notice I said “generally” and not “always”; there are circumstances where using asynchronous transformations can be beneficial and in this blog post I’ll demonstrate such a scenario, one that is pretty common when building data warehouses. Imagine I have a [Customer] dimension table that manages information about all of my customers as a slowly-changing dimension. If that is a type 2 slowly changing dimension then you will likely have multiple rows per customer in that table. Furthermore you might also have datetime fields that indicate the effective time period of each member record. Here is such a table that contains data for four dimension members {Terry, Max, Henry, Horace}: Notice that we have multiple records per customer and that the [SCDStartDate] of a record is equivalent to the [SCDEndDate] of the record that preceded it (if there was one). (Note that I am on record as saying I am not a fan of this technique of storing an [SCDEndDate] but for the purposes of clarity I have included it here.) Anyway, the idea here is that we will have some incoming data containing [CustomerName] & [EffectiveDate] and we need to use those values to lookup [Customer].[CustomerId]. The logic will be: Lookup a [CustomerId] WHERE [CustomerName]=[CustomerName] AND [SCDStartDate] <= [EffectiveDate] AND [EffectiveDate] <= [SCDEndDate] The conventional approach to this would be to use a full cached lookup but that isn’t an option here because we are using inequality conditions. The obvious next step then is to use a non-cached lookup which enables us to change the SQL statement to use inequality operators: Let’s take a look at the dataflow: Notice these are all synchronous components. This approach works just fine however it does have the limitation that it has to issue a SQL statement against your lookup set for every row thus we can expect the execution time of our dataflow to increase linearly in line with the number of rows in our dataflow; that’s not good. OK, that’s the obvious method. Let’s now look at a different way of achieving this using an asynchronous Merge Join transform coupled with a Conditional Split. I’ve shown it post-execution so that I can include the row counts which help to illustrate what is going on here: Notice that there are more rows output from our Merge Join component than on the input. That is because we are joining on [CustomerName] and, as we know, we have multiple records per [CustomerName] in our lookup set. Notice also that there are two asynchronous components in here (the Sort and the Merge Join). I have embedded a video below that compares the execution times for each of these two methods. The video is just over 8minutes long. View on Vimeo  For those that can’t be bothered watching the video I’ll tell you the results here. The dataflow that used the Lookup transform took 36 seconds whereas the dataflow that used the Merge Join took less than two seconds. An illustration in case it is needed: Pretty conclusive proof that in some scenarios it may be quicker to use an asynchronous component than a synchronous one. Your mileage may of course vary. The scenario outlined here is analogous to performance tuning procedural SQL that uses cursors. It is common to eliminate cursors by converting them to set-based operations and that is effectively what we have done here. Our non-cached lookup is performing a discrete operation for every single row of data, exactly like a cursor does. By eliminating this cursor-in-disguise we have dramatically sped up our dataflow. I hope all of that proves useful. You can download the package that I demonstrated in the video from my SkyDrive at http://cid-550f681dad532637.skydrive.live.com/self.aspx/Public/BlogShare/20100514/20100514%20Lookups%20and%20Merge%20Joins.zip Comments are welcome as always. @Jamiet Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!

    Read the article

  • Calling XAI Inbound Services from Oracle BI Publisher

    - by ACShorten
    Note: This technique requires Oracle BI Publisher 1.1.3.4.1 which supports Service Complex Types. Web Services require credentials for authentication. Note: The deafults for the product installation are used in this article. If your site uses alternative values then substitute those alternatives where applicable. Note: Examples shown in this article are examples for illustrative purposes only. When building a report in Oracle BI Publisher it may be necessary to call an XAI Inbound Service to get information via the object rather than directly calling the database tables for various reasons: The CLOB fields used in the Object are accessible for a report. Note: CLOB fields cannot be used as criteria in the current release. Objects can take advantage of algorithms to format or calculate additional data that is not stored in the database directly. For example, Information format strings can automatically generated by the object which gives consistent information between a report and the online screens. To use this facility the following process must be performed: Ensure that the product group, cisusers by default, is enabled for the SPLServiceBean in the console. This allows BI Publisher access to call Web Services directly. To ensure this follow the instructions below: Logon to the Oracle WebLogic server console using an appropriate administrator account. By default the user system or weblogic is provided for this purpose. Navigate to the Security Realms section and select your configured realm. This is set to myrealm by default. In the Roles and Policies section, expand the SPLService section of the Deployments option to reveal the SPLServiceBean roles. If there is no role associated with the SPLServiceBean, create a new EJB role and specify the cisusers role, by default. For example:   Add a Role Condition to the role just created, with a Predicate List of Group and specify cisusers as the Group Argument Name. For example: Save all your changes. The XAI Inbound Services to be used by BI Publisher must be defined prior to using the interface. Refer to the XAI Best Practices (Doc Id: 942074.1) from My Oracle Support or via the online help for more information about this process. Inside BI Publisher create your report, according to the BI Publisher documentation. When specifying the dataset, under the Data Model Report option, specify the following to use an XAI Inbound Service as a data source: Parameter Comment Type Web Service Complex Type true Username Any valid user name within the product. This user MUST have security access to the objects referenced in the XAI Inbound Service Password Authentication password for Username Timeout Timeout, in seconds, set for the Web Service call. For example 60 seconds. WSDL URL Use the WSDL URL on the XAI Inbound Service definition as your WSDL URL. It will be in the following format by default:http://<host>:<port>/<server>/XAIApp/xaiserver/<service>?WSDLwhere: <host> - Host Name of Web Application Server <port> - Port allocated to Web Application Server for product access <server> - Server context for server <service> - XAI Inbound Service Name Note: For customers using secure transmission should substitute https instead of http and use the HTTPS port allocated to the product at installation time. Web Service Select the name of the service that shows in the drop-down menu. If no service name shows up, it means that Publisher could not establish a connection with the server or WSDL name provided in the above URL in order to get the service name. See BI Publisher server log for more information. Method Select the name of the Method that shows in the drop-down menu. A method name should show in the Method drop-down menu once the Web Service name is selected. For example: Additionally, filters can be used from the Web Service that can be generated, required or optional, from the WSDL in the Parameter List. For example:

    Read the article

  • Validating Petabytes of Data with Regularity and Thoroughness

    - by rickramsey
    by Brian Zents When former Intel CEO Andy Grove said “only the paranoid survive,” he wasn’t necessarily talking about tape storage administrators, but it’s a lesson they’ve learned well. After all, tape storage is the last line of defense to prevent data loss, so tape administrators are extra cautious in making sure their data is secure. Not surprisingly, we are often asked for ways to validate tape media and the files on them. In the past, an administrator could validate the media, but doing so was often tedious or disruptive or both. The debut of the Data Integrity Validation (DIV) and Library Media Validation (LMV) features in the Oracle T10000C drive helped eliminate many of these pains. Also available with the Oracle T10000D drive, these features use hardware-assisted CRC checks that not only ensure the data is written correctly the first time, but also do so much more efficiently. Traditionally, a CRC check takes at least 25 seconds per 4GB file with a 2:1 compression ratio, but the T10000C/D drives can reduce the check to a maximum of nine seconds because the entire check is contained within the drive. No data needs to be sent to a host application. A time savings of at least 64 percent is extremely beneficial over the course of checking an entire 8.5TB T10000D tape. While the DIV and LMV features are better than anything else out there, what storage administrators really need is a way to check petabytes of data with regularity and thoroughness. With the launch of Oracle StorageTek Tape Analytics (STA) 2.0 in April, there is finally a solution that addresses this longstanding need. STA bundles these features into one interface to automate all media validation activities across all Oracle SL3000 and SL8500 tape libraries in an environment. And best of all, the validation process can be associated with the health checks an administrator would be doing already through STA. In fact, STA validates the media based on any of the following policies: Random Selection – Randomly selects media for validation whenever a validation drive in the standalone library or library complex is available. Media Health = Action – Selects media that have had a specified number of successive exchanges resulting in an Exchange Media Health of “Action.” You can specify from one to five exchanges. Media Health = Evaluate – Selects media that have had a specified number of successive exchanges resulting in an Exchange Media Health of “Evaluate.” You can specify from one to five exchanges. Media Health = Monitor – Selects media that have had a specified number of successive exchanges resulting in an Exchange Media Health of “Monitor.” You can specify from one to five exchanges. Extended Period of Non-Use – Selects media that have not had an exchange for a specified number of days. You can specify from 365 to 1,095 days (one to three years). Newly Entered – Selects media that have recently been entered into the library. Bad MIR Detected – Selects media with an exchange resulting in a “Bad MIR Detected” error. A bad media information record (MIR) indicates degraded high-speed access on the media. To avoid disrupting host operations, an administrator designates certain drives for media validation operations. If a host requests a file from media currently being validated, the host’s request takes priority. To ensure that the administrator really knows it is the media that is bad, as opposed to the drive, STA includes drive calibration and qualification features. In addition, validation requests can be re-prioritized or cancelled as needed. To ensure that a specific tape isn’t validated too often, STA prevents a tape from being validated twice within 24 hours via one of the policies described above. A tape can be validated more often if the administrator manually initiates the validation. When the validations are complete, STA reports the results. STA does not report simply a “good” or “bad” status. It also reports if media is even degraded so the administrator can migrate the data before there is a true failure. From that point, the administrators’ paranoia is relieved, as they have the necessary information to make a sound decision about the health of the tapes in their environment. About the Photograph Photograph taken by Rick Ramsey in Death Valley, California, May 2014 - Brian Follow OTN Garage on: Web | Facebook | Twitter | YouTube

    Read the article

  • Speeding up procedural texture generation

    - by FalconNL
    Recently I've begun working on a game that takes place in a procedurally generated solar system. After a bit of a learning curve (having neither worked with Scala, OpenGL 2 ES or Libgdx before), I have a basic tech demo going where you spin around a single procedurally textured planet: The problem I'm running into is the performance of the texture generation. A quick overview of what I'm doing: a planet is a cube that has been deformed to a sphere. To each side, a n x n (e.g. 256 x 256) texture is applied, which are bundled in one 8n x n texture that is sent to the fragment shader. The last two spaces are not used, they're only there to make sure the width is a power of 2. The texture is currently generated on the CPU, using the updated 2012 version of the simplex noise algorithm linked to in the paper 'Simplex noise demystified'. The scene I'm using to test the algorithm contains two spheres: the planet and the background. Both use a greyscale texture consisting of six octaves of 3D simplex noise, so for example if we choose 128x128 as the texture size there are 128 x 128 x 6 x 2 x 6 = about 1.2 million calls to the noise function. The closest you will get to the planet is about what's shown in the screenshot and since the game's target resolution is 1280x720 that means I'd prefer to use 512x512 textures. Combine that with the fact the actual textures will of course be more complicated than basic noise (There will be a day and night texture, blended in the fragment shader based on sunlight, and a specular mask. I need noise for continents, terrain color variation, clouds, city lights, etc.) and we're looking at something like 512 x 512 x 6 x 3 x 15 = 70 million noise calls for the planet alone. In the final game, there will be activities when traveling between planets, so a wait of 5 or 10 seconds, possibly 20, would be acceptable since I can calculate the texture in the background while traveling, though obviously the faster the better. Getting back to our test scene, performance on my PC isn't too terrible, though still too slow considering the final result is going to be about 60 times worse: 128x128 : 0.1s 256x256 : 0.4s 512x512 : 1.7s This is after I moved all performance-critical code to Java, since trying to do so in Scala was a lot worse. Running this on my phone (a Samsung Galaxy S3), however, produces a more problematic result: 128x128 : 2s 256x256 : 7s 512x512 : 29s Already far too long, and that's not even factoring in the fact that it'll be minutes instead of seconds in the final version. Clearly something needs to be done. Personally, I see a few potential avenues, though I'm not particularly keen on any of them yet: Don't precalculate the textures, but let the fragment shader calculate everything. Probably not feasible, because at one point I had the background as a fullscreen quad with a pixel shader and I got about 1 fps on my phone. Use the GPU to render the texture once, store it and use the stored texture from then on. Upside: might be faster than doing it on the CPU since the GPU is supposed to be faster at floating point calculations. Downside: effects that cannot (easily) be expressed as functions of simplex noise (e.g. gas planet vortices, moon craters, etc.) are a lot more difficult to code in GLSL than in Scala/Java. Calculate a large amount of noise textures and ship them with the application. I'd like to avoid this if at all possible. Lower the resolution. Buys me a 4x performance gain, which isn't really enough plus I lose a lot of quality. Find a faster noise algorithm. If anyone has one I'm all ears, but simplex is already supposed to be faster than perlin. Adopt a pixel art style, allowing for lower resolution textures and fewer noise octaves. While I originally envisioned the game in this style, I've come to prefer the realistic approach. I'm doing something wrong and the performance should already be one or two orders of magnitude better. If this is the case, please let me know. If anyone has any suggestions, tips, workarounds, or other comments regarding this problem I'd love to hear them.

    Read the article

  • How to switch between views in android?

    - by aurezza
    I've tried several methods to switch between two views in my program. I've tried creating a new thread then have the view run for 5 seconds before creating intent to start my main activity. This is the code snippet from the said view class: mHelpThread = new Thread(){ @Override public void run(){ try { synchronized(this){ // Wait given period of time or exit on touch wait(5000); } } catch(InterruptedException ex){ } finish(); // Run next activity Intent intent = new Intent(Intent.ACTION_MAIN, null); intent.addCategory(Intent.CATEGORY_HOME); startActivity(intent); //stop(); } }; mHelpThread.start(); I can access the said view without error but it doesn't disappear after 5 seconds nor did it switched to main view when I even utilized an onTouchEvent() to detect touch on the screen of which it should have automatically closed. I've also tried adding a button on the said view to manually switch to main view: @Override public void onCreate(Bundle savedInstanceState) { super.onCreate(savedInstanceState); setContentView(R.layout.help); final HelpView helpView = this; final Button btnback = (Button) findViewById(R.id.back); btnback.setOnClickListener(new View.OnClickListener(){ public void onClick(View v) { Intent intent = new Intent(helpView, MainActivity.class); startActivity(intent); } }); } These codes worked, though, for creating a launcher for my program. So I thought that it would work the same if I added an option for help/rules(for the game) that would switch to another view. I've only since started using eclipse for android so pardon my lack of knowledge. Here is also the snippet from my manifest: <uses-sdk android:minSdkVersion="11" android:targetSdkVersion="15" /> <application android:icon="@drawable/ic_launcher" android:label="@string/app_name" android:theme="@style/AppTheme" > <activity android:name="MainActivity" android:label="@string/title_activity_main"> <intent-filter> <action android:name="android.intent.action.MAIN" /> <category android:name="android.intent.category.DEFAULT"/> </intent-filter> <intent-filter></intent-filter> </activity> <activity android:name="SplashScreen" android:theme="@style/Theme.Transparent"> <intent-filter> <action android:name="android.intent.action.MAIN"/> <category android:name="android.intent.category.LAUNCHER"/> </intent-filter> </activity> <activity android:name="HelpView" android:theme="@style/Theme.Transparent"> <intent-filter> <action android:name="android.intent.action.MAIN"/> <category android:name="android.intent.category.DEFAULT"/> </intent-filter> <intent-filter></intent-filter> </activity> </application>

    Read the article

  • LSI SAS 9240-8i on Ubuntu 12.04 Hangs on Modprobe

    - by Francois Stark
    I used the LSI 9240-8i card on a smaller Intel motherboard with no problems in Ubuntu, with ZFS. However, we rebuilt the server to allow for more disks, using the ASROCK X79 Extreme 11 motherboard. It has 7 PCIe slots, and a LSI 2008 on-board. At first I thought the LSI 9240, when plugged in to PCIe, clashed with the on-board LSI 2008. Every time I plugged in the LSI 9240, modprobe would hang. Then I completely disabled the on-board LSI 2008, and the problem persisted. Last night it booted perfectly ONCE - all LSI cards and connected disks visible... However, all subsequent reboots failed. Both LSI cards' bios scans appear and they both see the disks connected to them, but Ubuntu modprobe hangs. Some selected dmesg lines, with both LSI cards enabled: [ 190.752100] megasas: [ 0]waiting for 1 commands to complete [ 195.772071] megasas: [ 5]waiting for 1 commands to complete [ 200.792079] megasas: [10]waiting for 1 commands to complete [ 205.812078] megasas: [15]waiting for 1 commands to complete [ 210.832037] megasas: [20]waiting for 1 commands to complete [ 215.852077] megasas: [25]waiting for 1 commands to complete [ 220.872072] megasas: [30]waiting for 1 commands to complete [ 225.892078] megasas: [35]waiting for 1 commands to complete [ 230.912086] megasas: [40]waiting for 1 commands to complete [ 235.932075] megasas: [45]waiting for 1 commands to complete [ 240.306157] usb 2-1.5: USB disconnect, device number 7 [ 240.952076] megasas: [50]waiting for 1 commands to complete [ 240.960034] INFO: task modprobe:233 blocked for more than 120 seconds. [ 240.960055] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 240.960067] modprobe D ffffffff81806200 0 233 146 0x00000004 [ 240.960075] ffff880806ae3b48 0000000000000086 ffff880806ae3ae8 ffffffff8101adf3 [ 240.960083] ffff880806ae3fd8 ffff880806ae3fd8 ffff880806ae3fd8 0000000000013780 [ 240.960090] ffffffff81c0d020 ffff880806acae00 ffff880806ae3b58 ffff880808961720 [ 240.960096] Call Trace: [ 240.960107] [<ffffffff8101adf3>] ? native_sched_clock+0x13/0x80 [ 240.960116] [<ffffffff816579cf>] schedule+0x3f/0x60 [ 240.960137] [<ffffffffa00093f5>] megasas_issue_blocked_cmd+0x75/0xb0 [megaraid_sas] [ 240.960144] [<ffffffff8108aa50>] ? add_wait_queue+0x60/0x60 [ 240.960154] [<ffffffffa000a6c9>] megasas_get_seq_num+0xd9/0x260 [megaraid_sas] [ 240.960164] [<ffffffffa000ab31>] megasas_start_aen+0x31/0x60 [megaraid_sas] [ 240.960174] [<ffffffffa00136f1>] megasas_probe_one+0x69a/0x81c [megaraid_sas] [ 240.960182] [<ffffffff813345bc>] local_pci_probe+0x5c/0xd0 [ 240.960189] [<ffffffff81335e89>] __pci_device_probe+0xf9/0x100 [ 240.960197] [<ffffffff8130ce6a>] ? kobject_get+0x1a/0x30 [ 240.960205] [<ffffffff81335eca>] pci_device_probe+0x3a/0x60 [ 240.960212] [<ffffffff813f5278>] really_probe+0x68/0x190 [ 240.960217] [<ffffffff813f5505>] driver_probe_device+0x45/0x70 [ 240.960223] [<ffffffff813f55db>] __driver_attach+0xab/0xb0 [ 240.960227] [<ffffffff813f5530>] ? driver_probe_device+0x70/0x70 [ 240.960233] [<ffffffff813f5530>] ? driver_probe_device+0x70/0x70 [ 240.960237] [<ffffffff813f436c>] bus_for_each_dev+0x5c/0x90 [ 240.960243] [<ffffffff813f503e>] driver_attach+0x1e/0x20 [ 240.960248] [<ffffffff813f4c90>] bus_add_driver+0x1a0/0x270 [ 240.960255] [<ffffffffa001e000>] ? 0xffffffffa001dfff [ 240.960260] [<ffffffff813f5b46>] driver_register+0x76/0x140 [ 240.960266] [<ffffffffa001e000>] ? 0xffffffffa001dfff [ 240.960271] [<ffffffff81335b66>] __pci_register_driver+0x56/0xd0 [ 240.960277] [<ffffffffa001e000>] ? 0xffffffffa001dfff [ 240.960286] [<ffffffffa001e09e>] megasas_init+0x9e/0x1000 [megaraid_sas] [ 240.960294] [<ffffffff81002040>] do_one_initcall+0x40/0x180 [ 240.960301] [<ffffffff810a82fe>] sys_init_module+0xbe/0x230 [ 240.960307] [<ffffffff81661ec2>] system_call_fastpath+0x16/0x1b [ 240.960314] INFO: task scsi_scan_7:349 blocked for more than 120 seconds.

    Read the article

  • SQLAuthority News – Speaking at Southeast Asia SharePoint Conference 2013 – Singapore

    - by pinaldave
    Two years ago I spoke at Southeast Asia SharePoint Conference 2011, Singapore and I had a fantastic time to present to the Singapore audience. The session was very well received and lots of interest was generated. The event is back again this year and with much bigger scale. I will be presenting on SQL Server and Sharepoint subject at the conference. Session Details: Title: Performance in 60 Seconds – Database Tricks Every SharePoint Developer & Admin MUST Know Abstract: SharePoint Developers and System Administrators often come across situations where they face a slow server response, even though their hardware specifications are above  par. This session is for all the SharePoint Developers who want their server to perform at blazing fast speed but want to invest very little time to make it happen. We will go over various database tricks which require absolutely no time to master and require practically no SQL coding at all. After attending this session, Developers will only need 60 seconds to improve performance of their database server in their SharePoint implementation. Date and Time: January 18, 20013 - 3:15 PM-4:15 PM Location: Max Atria is located at Singapore Expo, 1 Expo Drive, Singapore Tel 65 6403 2160 This session will cover lots of interesting tips and tricks about SQL Server and SharePoint co-exists together. I promise that every attendee will walk out with a trick which they can walk out of session and directly apply to their production server to improve its performance. The event is going to be again fantastic event – if you are in Singapore – you must not miss this event. If you are planning vacation – this is the right time to take days off and travel to Singapore for vacation. The event features over 30 sessions to choose from, focus on three areas of business gain: Exploring Information, Improving Productivity and Making it Work. This event has an excellent line up of international speakers (speakers traveling from the USA, Australia, New Zealand, Sri Lanka and India). Register early to reserve a spot at your choice of more than 30 classes taught by Microsoft Certified Masters, MVPs, and other top SharePoint experts! Here I have attempted to answer a few of the questions which every SharePoint professional half: Which sessions suit my skill level? Click here. What sessions are right for me? Click here. Which sessions are of my interests? Click here. Which sessions are on when? Click here. If you register by next Friday, 14, December – you can save $126 on the regular price of the conference. Prizes, Giveaways and … I love conference goodies – I collect them as a souvenir . This event is known for its generous prizes. The first 100 people to register on the day will get a SPECIAL gift at the event. Additionally there are exhibitor booth give away too. Here is the page listing all the prizes and giveaways. Do leave a comment or send me email if you are going to the event, we can sit together and have a coffee. Reference: Pinal Dave (http://blog.sqlauthority.com) Filed under: PostADay, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, SQLAuthority News, T SQL, Technology

    Read the article

  • How does interpolation actually work to smooth out an object's movement?

    - by user22241
    I've asked a few similar questions over the past 8 months or so with no real joy, so I am going make the question more general. I have an Android game which is OpenGL ES 2.0. within it I have the following Game Loop: My loop works on a fixed time step principle (dt = 1 / ticksPerSecond) loops=0; while(System.currentTimeMillis() > nextGameTick && loops < maxFrameskip){ updateLogic(dt); nextGameTick+=skipTicks; timeCorrection += (1000d/ticksPerSecond) % 1; nextGameTick+=timeCorrection; timeCorrection %=1; loops++; } render(); My intergration works like this: sprite.posX+=sprite.xVel*dt; sprite.posXDrawAt=sprite.posX*width; Now, everything works pretty much as I would like. I can specify that I would like an object to move across a certain distance (screen width say) in 2.5 seconds and it will do just that. Also because of the frame skipping that I allow in my game loop, I can do this on pretty much any device and it will always take 2.5 seconds. Problem However, the problem is that when a render frame skips, the graphic stutter. It's extremely annoying. If I remove the ability to skip frames, then everything is smooth as you like, but will run at different speeds on different devices. So it's not an option. I'm still not sure why the frame skips, but I would like to point out that this is Nothing to do with poor performance, I've taken the code right back to 1 tiny sprite and no logic (apart from the logic required to move the sprite) and I still get skipped frames. And this is on a Google Nexus 10 tablet (and as mentioned above, I need frame skipping to keep the speed consistent across devices anyway). So, the only other option I have is to use interpolation (or extrapolation), I've read every article there is out there but none have really helped me to understand how it works and all of my attempted implementations have failed. Using one method I was able to get things moving smoothly but it was unworkable because it messed up my collision. I can foresee the same issue with any similar method because the interpolation is passed to (and acted upon within) the rendering method - at render time. So if Collision corrects position (character now standing right next to wall), then the renderer can alter it's position and draw it in the wall. So I'm really confused. People have said that you should never alter an object's position from within the rendering method, but all of the examples online show this. So I'm asking for a push in the right direction, please do not link to the popular game loop articles (deWitters, Fix your timestep, etc) as I've read these multiple times. I'm not asking anyone to write my code for me. Just explain please in simple terms how Interpolation actually works with some examples. I will then go and try to integrate any ideas into my code and will ask more specific questions if need-be further down the line. (I'm sure this is a problem many people struggle with).

    Read the article

  • New Analytic settings for the new code

    - by Steve Tunstall
    If you have upgraded to the new 2011.1.3.0 code, you may find some very useful settings for the Analytics. If you didn't already know, the analytic datasets have the potential to fill up your OS hard drives. The more datasets you use and create, that faster this can happen. Since they take a measurement every second, forever, some of these metrics can get in the multiple GB size in a matter of weeks. The traditional 'fix' was that you had to go into Analytics -> Datasets about once a month and clean up the largest datasets. You did this by deleting them. Ouch. Now you lost all of that historical data that you might have wanted to check out many months from now. Or, you had to export each metric individually to a CSV file first. Not very easy or fun. You could also suspend a dataset, and have it not collect data at all. Well, that fixed the problem, didn't it? of course you now had no data to go look at. Hmmmm.... All of this is no longer a concern. Check out the new Settings tab under Analytics... Now, I can tell the ZFSSA to keep every second of data for, say, 2 weeks, and then average those 60 seconds of each minute into a single 'minute' value. I can go even further and ask it to average those 60 minutes of data into a single 'hour' value.  This allows me to effectively shrink my older datasets by a factor of 1/3600 !!! Very cool. I can now allow my datasets to go forever, and really never have to worry about them filling up my OS drives. That's great going forward, but what about those huge datasets you already have? No problem. Another new feature in 2011.1.3.0 is the ability to shrink the older datasets in the same way. Check this out. I have here a dataset called "Disk: I/O opps per second" that is about 6.32M on disk (You need not worry so much about the "In Core" value, as that is in RAM, and it fluctuates all the time. Once you stop viewing a particular metric, you will see that shrink over time, just relax).  When one clicks on the trash can icon to the right of the dataset, it used to delete the whole thing, and you would have to re-create it from scratch to get the data collecting again. Now, however, it gives you this prompt: As you can see, this allows you to once again shrink the dataset by averaging the second data into minutes or hours. Here is my new dataset size after I do this. So it shrank from 6.32MB down to 2.87MB, but i can still see my metrics going back to the time I began the dataset. Now, you do understand that once you do this, as you look back in time to the minute or hour data metrics, that you are going to see much larger time values, right? You will need to decide what size of granularity you can live with, and for how long. Check this out. Here is my Disk: Percent utilized from 5-21-2012 2:42 pm to 4:22 pm: After I went through the delete process to change everything older than 1 week to "Minutes", the same date and time looks like this: Just understand what this will do and how you want to use it. Right now, I'm thinking of keeping the last 6 weeks of data as "seconds", and then the last 3 months as "Minutes", and then "Hours" forever after that. I'll check back in six months and see how the sizes look. Steve 

    Read the article

  • World Record Siebel PSPP Benchmark on SPARC T4 Servers

    - by Brian
    Oracle's SPARC T4 servers set a new World Record for Oracle's Siebel Platform Sizing and Performance Program (PSPP) benchmark suite. The result used Oracle's Siebel Customer Relationship Management (CRM) Industry Applications Release 8.1.1.4 and Oracle Database 11g Release 2 running Oracle Solaris on three SPARC T4-2 and two SPARC T4-1 servers. The SPARC T4 servers running the Siebel PSPP 8.1.1.4 workload which includes Siebel Call Center and Order Management System demonstrates impressive throughput performance of the SPARC T4 processor by achieving 29,000 users. This is the first Siebel PSPP 8.1.1.4 benchmark supporting 29,000 concurrent users with a rate of 239,748 Business Transactions/hour. The benchmark demonstrates vertical and horizontal scalability of Siebel CRM Release 8.1.1.4 on SPARC T4 servers. Performance Landscape Systems Txn/hr Users Call Center Order Management Response Times (sec) 1 x SPARC T4-1 (1 x SPARC T4 2.85 GHz) – Web 3 x SPARC T4-2 (2 x SPARC T4 2.85 GHz) – App/Gateway 1 x SPARC T4-1 (1 x SPARC T4 2.85 GHz) – DB 239,748 29,000 0.165 0.925 Oracle: Call Center + Order Management Transactions: 197,128 + 42,620 Users: 20300 + 8700 Configuration Summary Web Server Configuration: 1 x SPARC T4-1 server 1 x SPARC T4 processor, 2.85 GHz 128 GB memory Oracle Solaris 10 8/11 iPlanet Web Server 7 Application Server Configuration: 3 x SPARC T4-2 servers, each with 2 x SPARC T4 processor, 2.85 GHz 256 GB memory 3 x 300 GB SAS internal disks Oracle Solaris 10 8/11 Siebel CRM 8.1.1.5 SIA Database Server Configuration: 1 x SPARC T4-1 server 1 x SPARC T4 processor, 2.85 GHz 128 GB memory Oracle Solaris 11 11/11 Oracle Database 11g Release 2 (11.2.0.2) Storage Configuration: 1 x Sun Storage F5100 Flash Array 80 x 24 GB flash modules Benchmark Description Siebel 8.1 PSPP benchmark includes Call Center and Order Management: Siebel Financial Services Call Center – Provides the most complete solution for sales and service, allowing customer service and telesales representatives to provide superior customer support, improve customer loyalty, and increase revenues through cross-selling and up-selling. High-level description of the use cases tested: Incoming Call Creates Opportunity, Quote and Order and Incoming Call Creates Service Request . Three complex business transactions are executed simultaneously for specific number of concurrent users. The ratios of these 3 scenarios were 30%, 40%, 30% respectively, which together were totaling 70% of all transactions simulated in this benchmark. Between each user operation and the next one, the think time averaged approximately 10, 13, and 35 seconds respectively. Siebel Order Management – Oracle's Siebel Order Management allows employees such as salespeople and call center agents to create and manage quotes and orders through their entire life cycle. Siebel Order Management can be tightly integrated with back-office applications allowing users to perform tasks such as checking credit, confirming availability, and monitoring the fulfillment process. High-level description of the use cases tested: Order & Order Items Creation and Order Updates. Two complex Order Management transactions were executed simultaneously for specific number of concurrent users concurrently with aforementioned three Call Center scenarios above. The ratio of these 2 scenarios was 50% each, which together were totaling 30% of all transactions simulated in this benchmark. Between each user operation and the next one, the think time averaged approximately 20 and 67 seconds respectively. Key Points and Best Practices No processor cores or cache were activated or deactivated on the SPARC T-Series systems to achieve special benchmark effects. See Also Siebel White Papers SPARC T4-1 Server oracle.com OTN SPARC T4-2 Server oracle.com OTN Siebel CRM oracle.com OTN Oracle Solaris oracle.com OTN Oracle Database 11g Release 2 Enterprise Edition oracle.com OTN Disclosure Statement Copyright 2012, Oracle and/or its affiliates. All rights reserved. Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names may be trademarks of their respective owners. Results as of 30 September 2012.

    Read the article

  • The Best Data Integration for Exadata Comes from Oracle

    - by maria costanzo
    Oracle Data Integrator and Oracle GoldenGate offer unique and optimized data integration solutions for Oracle Exadata. For example, customers that choose to feed their data warehouse or reporting database with near real-time throughout the day, can do so without decreasing  performance or availability of source and target systems. And if you ask why real-time, the short answer is: in today’s fast-paced, always-on world, business decisions need to use more relevant, timely data to be able to act fast and seize opportunities. A longer response to "why real-time" question can be found in a related blog post. If we look at the solution architecture, as shown on the diagram below,  Oracle Data Integrator and Oracle GoldenGate are both uniquely designed to take full advantage of the power of the database and to eliminate unnecessary middle-tier components. Oracle Data Integrator (ODI) is the best bulk data loading solution for Exadata. ODI is the only ETL platform that can leverage the full power of Exadata, integrate directly on the Exadata machine without any additional hardware, and by far provides the simplest setup and fastest overall performance on an Exadata system. We regularly see customers achieving a 5-10 times boost when they move their ETL to ODI on Exadata. For  some companies the performance gain is even much higher. For example a large insurance company did a proof of concept comparing ODI vs a traditional ETL tool (one of the market leaders) on Exadata. The same process that was taking 5hrs and 11 minutes to complete using the competing ETL product took 7 minutes and 20 seconds with ODI. Oracle Data Integrator was 42 times faster than the conventional ETL when running on Exadata.This shows that Oracle's own data integration offering helps you to gain the most out of your Exadata investment with a truly optimized solution. GoldenGate is the best solution for streaming data from heterogeneous sources into Exadata in real time. Oracle GoldenGate can also be used together with Data Integrator for hybrid use cases that also demand non-invasive capture, high-speed real time replication. Oracle GoldenGate enables real-time data feeds from heterogeneous sources non-invasively, and delivers to the staging area on the target Exadata system. ODI runs directly on Exadata to use the database engine power to perform in-database transformations. Enterprise Data Quality is integrated with Oracle Data integrator and enables ODI to load trusted data into the data warehouse tables. Only Oracle can offer all these technical benefits wrapped into a single intelligence data warehouse solution that runs on Exadata. Compared to traditional ETL with add-on CDC this solution offers: §  Non-invasive data capture from heterogeneous sources and avoids any performance impact on source §  No mid-tier; set based transformations use database power §  Mini-batches throughout the day –or- bulk processing nightly which means maximum availability for the DW §  Integrated solution with Enterprise Data Quality enables leveraging trusted data in the data warehouse In addition to Starwood Hotels and Resorts, Morrison Supermarkets, United Kingdom’s fourth-largest food retailer, has seen the power of this solution for their new BI platform and shared their story with us. Morrisons needed to analyze data across a large number of manufacturing, warehousing, retail, and financial applications with the goal to achieve single view into operations for improved customer service. The retailer deployed Oracle GoldenGate and Oracle Data Integrator to bring new data into Oracle Exadata in near real-time and replicate the data into reporting structures within the data warehouse—extending visibility into operations. Using Oracle's data integration offering for Exadata, Morrisons produced financial reports in seconds, rather than minutes, and improved staff productivity and agility. You can read more about Morrison’s success story here and hear from Starwood here. From an Irem Radzik article.

    Read the article

  • Graphics trouble after resuming from hibernate or suspend

    - by Voyagerfan5761
    I have a Dell Inspiron 2650 (with NVidia graphics, using nouveau drivers) that I'm using to try out Ubuntu. It's all great, except that Hibernate and Suspend aren't usable. Yes, I know that questions about power-save issues are rampant in the Linux support universe, but it seems that every time I find a solution it's for a very specific hardware combination and doesn't apply to me. So anyway, here goes. When I resume from either power-saving mode, I'll get graphics problems anywhere on the range from a few scattered random-colored pixels that won't change; all the way to full-screen patterns that don't change as I move the mouse, hit keys on the keyboard, or even bring up the shutdown dialog using the power button. Those full-screen issues (which may involve stripes with random pixels, partial black screen, or both) always end in me forcing the machine to shut down by holding the power button. I haven't done much testing yet to determine what severity level is most commonly associated with each mode, but I do avoid using either power-save option because of these issues. I'll add info on my hardware as I can gather it (no home internet connection, and this laptop is tethered to my desk by a dead battery and casing degradation). Please feel free to request something specific in the question comments. Hardware Info See this hardinfo report for my system's hardware configuration. (No, my username is not "myuser"; I sanitized hardinfo's output before publishing it.) Screenshots These screenshots are from a relatively mild occurrence, which happened after the second hibernation I took that session. The first one worked great, though I used the wireless card and Firefox heavily between the two hibernation attempts. Take a look at what happened when I opened my home directory in Nautilus and scrolled it: See below for the situations I've tested so far. The real trouble comes when the machine resumes to an unusable state; in such cases I can't even unlock the screen or properly reboot, much less take a screenshot. I have a hunch that putting a CD in the drive will cause such major failures, and I will try that at some point; see related question. Situations Tested Maverick (10.10) Suspend Seems to suspend nicely with nothing running Seems to suspend nicely with flash drive plugged in On resume from suspend with no flash drive, Terminal and gedit running: Funky graphics on top of log output, then blank screen with pixelated cursor; no response to power button (normally will shutdown 60 seconds later) Hibernate Seems to hibernate nicely with nothing running Seems to hibernate nicely with a few apps (Terminal, Mouse preferences) running Seems to not hibernate when flash drive plugged in Seems to not hibernate when System Monitor is running Have encountered failed hibernation (after several hours and one successful hibernate/thaw cycle) with no external media connected and no programs running except normal background stuff Natty LiveCD (11.04_2010-12-22) When I tested it, Natty wouldn't stay logged in. It played part of the login sound and then [ OK ] appeared in the top right corner (white-on-black terminal text) for a few seconds. Then it kicked me back to the Unlock screen. It did that four times before I gave up and just tested suspend from the Unlock screen. Suspend Resumed to vertical gray and black lines 2px (?) wide, then shifted to vertical "jail bars" of black over a black screen with above-described random pixels and mouse pointer. No apparent response to input from mouse (clicking randomly). Keyboard and touchpad unrecognized.

    Read the article

  • PHP-FPM stops responding and dies [migrated]

    - by user12361
    I'm running Drupal 6 with Nginx 1.5.1 and PHP-FPM (PHP 5.3.26) on a 1GB single core VPS with 3GB of swap space on SSD storage. I just switched from shared hosting to this unmanaged VPS because my site was getting too heavy, so I'm still learning the ropes. I have moderately high traffic, I don't really monitor it closely but Google Adsense usually record close to 30K page views/day. I usually have 50 to 80 authenticated users logged in and a few hundred more anonymous users hitting the Boost static HTML cache at any given moment. The problem I'm having is that PHP-FPM frequently stops responding, resulting in Nginx 502 or 504 errors. I swear I have read every page on the internet about this issue, which seems fairly common, and I've tried endless combinations of configurations, and I can't find a good solution. After restarting Nginx and PHP-FPM, the site runs really fast for a while, and then without warning it simply stops responding. I get a white screen while the browser waits on the server, and after about 30 seconds to a minute it throws an Nginx 502 or 504 error. Sometimes it runs well for 2 minutes, sometimes 5 minutes, sometimes 5 hours, but it always ends up hanging. When I find the server in this state, there is still plenty of free memory (500MB or more) and no major CPU usage, the control and worker PHP-FPM processes are still present, and the server is still pingable and usable via SSH. A reload of PHP-FPM via the init script revives it again. The hangups don't seem to correspond to the amount of traffic, because I observed this behavior consistently when I was testing this configuration on a development VPS with no traffic at all. I've been constantly tweaking the settings, but I can't definitively eliminate the problem. I set Nginx workers to just 1. In the PHP-FPM config I have tried all three of the process managers. "Dynamic" is definitely the least reliable, consistently hanging up after only a few minutes. "Static" also has been unreliable and unpredictable. The least buggy has been "ondemand", but even that is failing me, sometimes after as much as 12 to 24 hours. But I can't leave the server unattended because PHP-FPM dies and never comes back on its own. I tried adjusting the pm.max_children value from as low as 3 to as high as 50, doesn't make a lot of difference, but I currently have it at 10. Same thing for the spare servers values. I also have set pm.max_requests anywhere from 30 to unlimited, and it doesn't seem to make a difference. According to the logs, the PHP-FPM processes are not exiting with SIGSEGV or SIGBUS, but rather with SIGTERM. I get a lot of lines like: WARNING: [pool www] child 3739, script '/var/www/drupal6/index.php' (request: "GET /index.php") execution timed out (38.739494 sec), terminating and: WARNING: [pool www] child 3738 exited on signal 15 (SIGTERM) after 50.004380 seconds from start I actually found several articles that recommend doing a graceful reload of PHP-FPM via cron every few minutes or hours to circumvent this issue. So that's what I did, "/etc/init.d/php-fpm reload" every 5 minutes. So far, it's keeping the lights on. But it feels like a dreadful hack. Is PHP-FPM really that unreliable? Is there anything else I can do? Thanks a lot!

    Read the article

  • Why is the framerate (fps) capped at 60?

    - by dennmat
    ISSUE I recently moved a project from my laptop to my desktop(machine info below). On my laptop the exact same code displays the fps(and ms/f) correctly. On my desktop it does not. What I mean by this is on the laptop it will display 300 fps(for example) where on my desktop it will show only up to 60. If I add 100 objects to the game on the laptop I'll see my frame rate drop accordingly; the same test on the desktop results in no change and the frames stay at 60. It takes a lot(~300) entities before I'll see a frame drop on the desktop, then it will descend. It seems as though its "theoretical" frames would be 400 or 500 but will never actually get to that and only do 60 until there's too much to handle at 60. This 60 frame cap is coming from no where. I'm not doing any frame limiting myself. It seems like something external is limiting my loop iterations on the desktop, but for the last couple days I've been scratching my head trying to figure out how to debug this. SETUPS Desktop: Visual Studio Express 2012 Windows 7 Ultimate 64-bit Laptop: Visual Studio Express 2010 Windows 7 Ultimate 64-bit The libraries(allegro, box2d) are the same versions on both setups. CODE Main Loop: while(!abort) { frameTime = al_get_time(); if (frameTime - lastTime >= 1.0) { lastFps = fps/(frameTime - lastTime); lastTime = frameTime; avgMspf = cumMspf/fps; cumMspf = 0.0; fps = 0; } /** DRAWING/UPDATE CODE **/ fps++; cumMspf += al_get_time() - frameTime; } Note: There is no blocking code in the loop at any point. Where I'm at My understanding of al_get_time() is that it can return different resolutions depending on the system. However the resolution is never worse than seconds, and the double is represented as [seconds].[finer-resolution] and seeing as I'm only checking for a whole second al_get_time() shouldn't be responsible. My project settings and compiler options are the same. And I promise its the same code on both machines. My googling really didn't help me much, and although technically it's not that big of a deal. I'd really like to figure this out or perhaps have it explained, whichever comes first. Even just an idea of how to go about figuring out possible causes, because I'm out of ideas. Any help at all is greatly appreciated. EDIT: Thanks All. For any others that find this to disable vSync(windows only) in opengl: First get "wglext.h". It's all over the web. Then you can use a tool like GLee or just write your own quick extensions manager like: bool WGLExtensionSupported(const char *extension_name) { PFNWGLGETEXTENSIONSSTRINGEXTPROC _wglGetExtensionsStringEXT = NULL; _wglGetExtensionsStringEXT = (PFNWGLGETEXTENSIONSSTRINGEXTPROC) wglGetProcAddress("wglGetExtensionsStringEXT"); if (strstr(_wglGetExtensionsStringEXT(), extension_name) == NULL) { return false; } return true; } and then create and setup your function pointers: PFNWGLSWAPINTERVALEXTPROC wglSwapIntervalEXT = NULL; PFNWGLGETSWAPINTERVALEXTPROC wglGetSwapIntervalEXT = NULL; if (WGLExtensionSupported("WGL_EXT_swap_control")) { // Extension is supported, init pointers. wglSwapIntervalEXT = (PFNWGLSWAPINTERVALEXTPROC) wglGetProcAddress("wglSwapIntervalEXT"); // this is another function from WGL_EXT_swap_control extension wglGetSwapIntervalEXT = (PFNWGLGETSWAPINTERVALEXTPROC) wglGetProcAddress("wglGetSwapIntervalEXT"); } Then just call wglSwapIntervalEXT(0) to disable vSync and 1 to enable vSync. I found the reason this is windows only is that openGl actually doesn't deal with anything other than rendering it leaves the rest up to the OS and Hardware. Thanks everyone saved me a lot of time!

    Read the article

  • 12.10 unable to install or even run from Live CD with nVidia GTX 580

    - by user99056
    I've used Ubuntu in the past (set up as web server, etc over in Iraq), so I'm not a 100% Linux Noob, however, I'm running into a brick wall here. I've got a machine I built when I got back to the US earlier this year, running Windows 7 Ultimate on it, and I've now got some free time and would like to transition over to Ubuntu full time. I've searched around in the forums, and there seems to be an issue with the nVidia graphics cards, so I've tried going to the EVGA site to see if I could find a new BIOS update for it and had no luck, so I'm back searching the forums here again and decided to just go ahead and post my question. My apologies if this is covered in another post and I was just unable to find it. I've found a few 'similar' posts, but nothing as bad as my issue. With the history aside, here is the actual detailed issue: I purchased a new SSD (Intel 520 SSD), arrived today, and I disconnect my old Windows 7 SSD. I had pre downloaded the ubuntu-12.10-desktop-amd64 earlier today and burned it to DVD. Upon inserting the Live CD into the computer and booting up, everything was fine up to the 'Run From Live CD' or 'Install Ubuntu Now' buttons. As I was sure I wanted to go ahead and make the switch, I selected the 'Install Now' from the right hand side. CD Spins up, black window pops up, and then the errors started: date/time GPU Lockup date/time Failed to idle channel 1 date/time PFIFO - playlist update failed date/time Failed to idle channel 2 date/time PFIFO - playlist update failed Thinking it might correct itself, I let it run and it would swap over to a GUI Screen that was locked up with major blurring/etc, then back to the command line with the errors. Eventually it said something along the lines of 'unknown status' and switched back to the GUI and froze. So, that's when I tried to see if I could find a BIOS upgrade for the nVidia GTX580 cards, and had no luck. So I thought, why not try to just run it from the Live CD and see if I can at least get a look at it, maybe if I could get it running try to do some sort of install from there and fix the driver issue. I rebooted, brought up the Live CD, and this time chose the left option / run from the CD. It brought me all the way in to the desktop, I saw my drives, the other icons, could move the mouse, etc for about 30 seconds and then it locked up completely. I've tried this a couple of times and get the same results every time. Hardware: Intel i7-3930K CPU @ 3.2GHz (12 CPUs) / MSI MS-7760 Motherboard / 32GB RAM / 2 x EVGA (nVidia) GeForce GTX 580 (4GB Ram each) So the question is: Is there any way to install 12.10 if you can't even get the Live CD to run (for more than 30 seconds)? My current hardware configuration is both of the GTX 580 cards have an SLI jumper on them, and I have 2 monitors on each card. (Ubuntu info obviously only shows on the main monitor from the failed installation and the attempt at running the Live CD). Perhaps opening the machine back up and removing the SLI Jumper and removing the other 3 monitors (so it only would have 1 video card with one monitor on it) would actually allow me to get 12.10 installed, then I could work on an nVidia Video Driver fix for the GTX 580, and then possibly hook up the other video card and monitors? Or is this something that they are currently aware of and may update with a future release in the next few days/weeks? Any thoughts or suggestions would be greatly appreciated, as I can't even try to fix the issue (assuming it is the nVidia drivers) if I can't even get it to install at all.

    Read the article

  • How to improve WinForms MSChart performance?

    - by Marcel
    Hi all, I have created some simple charts (of type FastLine) with MSChart and update them with live data, like below: . To do so, I bind an observable collection of a custom type to the chart like so: // set chart data source this._Chart.DataSource = value; //is of type ObservableCollection<SpectrumLevels> //define x and y value members for each series this._Chart.Series[0].XValueMember = "Index"; this._Chart.Series[1].XValueMember = "Index"; this._Chart.Series[0].YValueMembers = "Channel0Level"; this._Chart.Series[1].YValueMembers = "Channel1Level"; // bind data to chart this._Chart.DataBind(); //lasts 1.5 seconds for 8000 points per series At each refresh, the dataset completely changes, it is not a scrolling update! With a profiler I have found that the DataBind() call takes about 1.5 seconds. The other calls are negligible. How can I make this faster? Should I use another type than ObservableCollection? An array probably? Should I use another form of data binding? Is there some tweak for the MSChart that I may have missed? Should I use a sparsed set of date, having one value per pixel only? Have I simply reached the performance limit of MSCharts? From the type of the application to keep it "fluent", we should have multiple refreshes per second. Thanks for any hints!

    Read the article

  • Algorithm design, "randomising" timetable schedule in Python although open to other languages.

    - by S1syphus
    Before I start I should add I am a musician and not a native programmer, this was undertook to make my life easier. Here is the situation, at work I'm given a new csv file each which contains a list of sound files, their length, and the minimum total amount of time they must be played. I create a playlist of exactly 60 minutes, from this excel file. Each sample played the by the minimum number of instances, but spread out from each other; so there will never be a period where for where one sound is played twice in a row or in close proximity to itself. Secondly, if the minimum instances of each song has been used, and there is still time with in the 60 min, it needs to fill the remaining time using sounds till 60 minutes is reached, while adhering to above. The smallest duration possible is 15 seconds, and then multiples of 15 seconds. Here is what I came up with in python and the problems I'm having with it, and as one user said its buggy due to the random library used in it. So I'm guessing a total rethink is on the table, here is where I need your help. Whats is the best way to solve the issue, I have had a brief look at things like knapsack and bin packing algorithms, while both are relevant neither are appropriate and maybe a bit beyond me.

    Read the article

< Previous Page | 58 59 60 61 62 63 64 65 66 67 68 69  | Next Page >