Search Results

Search found 5454 results on 219 pages for 'soa purge instances dehyd'.

Page 21/219 | < Previous Page | 17 18 19 20 21 22 23 24 25 26 27 28  | Next Page >

  • Execute .sh script on ec2 instances without rebooting

    - by waigani
    I currently keep my app code on S3 and have a startup.sh script which is fired via /etc/rc.local and installs the apps and any edits etc. Thus when I make a change, I need to reboot all my instances for the change to take effect. Is there a way to trigger the script without rebooting the instance? EDIT: I do not want to individually log into all my instances. I would prefer a method that I can script up to apply to all my instances at once - which are in an autoscaling group.

    Read the article

  • Windows Azure Use Case: Hybrid Applications

    - by BuckWoody
    This is one in a series of posts on when and where to use a distributed architecture design in your organization's computing needs. You can find the main post here: http://blogs.msdn.com/b/buckwoody/archive/2011/01/18/windows-azure-and-sql-azure-use-cases.aspx  Description: Organizations see the need for computing infrastructures that they can “rent” or pay for only when they need them. They also understand the benefits of distributed computing, but do not want to create this infrastructure themselves. However, they may have considerations that prevent them from moving all of their current IT investment to a distributed environment: Private data (do not want to send or store sensitive data off-site) High dollar investment in current infrastructure Applications currently running well, but may need additional periodic capacity Current applications not designed in a stateless fashion In these situations, a “hybrid” approach works best. In fact, with Windows Azure, a hybrid approach is an optimal way to implement distributed computing even when the stipulations above do not apply. Keeping a majority of the computing function in an organization local while exploring and expanding that footprint into Windows and SQL Azure is a good migration or expansion strategy. A “hybrid” architecture merely means that part of a computing cycle is shared between two architectures. For instance, some level of computing might be done in a Windows Azure web-based application, while the data is stored locally at the organization. Implementation: There are multiple methods for implementing a hybrid architecture, in a spectrum from very little interaction from the local infrastructure to Windows or SQL Azure. The patterns fall into two broad schemas, and even these can be mixed. 1. Client-Centric Hybrid Patterns In this pattern, programs are coded such that the client system sends queries or compute requests to multiple systems. The “client” in this case might be a web-based codeset actually stored on another system (which acts as a client, the user’s device serving as the presentation layer) or a compiled program. In either case, the code on the client requestor carries the burden of defining the layout of the requests. While this pattern is often the easiest to code, it’s the most brittle. Any change in the architecture must be reflected on each client, but this can be mitigated by using a centralized system as the client such as in the web scenario. 2. System-Centric Hybrid Patterns Another approach is to create a distributed architecture by turning on-site systems into “services” that can be called from Windows Azure using the service Bus or the Access Control Services (ACS) capabilities. Code calls from a series of in-process client application. In this pattern you move the “client” interface into the server application logic. If you do not wish to change the application itself, you can “layer” the results of the code return using a product (such as Microsoft BizTalk) that exposes a Web Services Definition Language (WSDL) endpoint to Windows Azure using the Application Fabric. In effect, this is similar to creating a Service Oriented Architecture (SOA) environment, and has the advantage of de-coupling your computing architecture. If each system offers a “service” of the results of some software processing, the operating system or platform becomes immaterial, assuming it adheres to a service contract. There are important considerations when you federate a system, whether to Windows or SQL Azure or any other distributed architecture. While these considerations are consistent with coding any application for distributed computing, they are especially important for a hybrid application. Connection resiliency - Applications on-premise normally have low-latency and good connection properties, something you’re not always guaranteed in a distributed and hybrid application. Whether a centralized client or a distributed one, the code should be able to handle extended retry logic. Authorization and Access - In a single authorization environment like a Active Directory domain, security is handled at a user-password level. In a distributed computing environment, you have more options. You can mitigate this with  using The Windows Azure Application Fabric feature of ACS to make the Azure application aware of the App Fabric as an ADFS provider. However, a claims-based authentication structure is often a superior choice.  Consistency and Concurrency - When you have a Relational Database Management System (RDBMS), Consistency and Concurrency are part of the design. In a Service Architecture, you need to plan for sequential message handling and lifecycle. Resources: How to Build a Hybrid On-Premise/In Cloud Application: http://blogs.msdn.com/b/ignitionshowcase/archive/2010/11/09/how-to-build-a-hybrid-on-premise-in-cloud-application.aspx  General Architecture guidance: http://blogs.msdn.com/b/buckwoody/archive/2010/12/21/windows-azure-learning-plan-architecture.aspx   

    Read the article

  • List of all states from COMPOSITE_INSTANCE, CUBE_INSTANCE, DLV_MESSAGE tables

    - by Deepak Arora
    In many of my engagements I get asked repeatedly about the states of the composites in 11g and how to decipher them, especially when we are troubleshooting issues around purging. I have compiled a list of all the states from the COMPOSITE_INSTANCE, CUBE_INSTANCE, and DLV_MESSAGE tables. These are the primary tables that are used when using BPEL composites and how they are used with the ECID.  Composite State Values COMPOSITE_INSTANCE States State Description 0 Running 1 Completed 2 Running with faults 3 Completed with faults 4 Running with recovery required 5 Completed with recovery required 6 Running with faults and recovery required 7 Completed with faults and recovery required 8 Running with suspended 9 Completed with suspended 10 Running with faults and suspended 11 Completed with faults and suspended 12 Running with recovery required and suspended 13 Completed with recovery required and suspended 14 Running with faults, recovery required, and suspended 15 Completed with faults, recovery required, and suspended 16 Running with terminated 17 Completed with terminated 18 Running with faults and terminated 19 Completed with faults and terminated 20 Running with recovery required and terminated 21 Completed with recovery required and terminated 22 Running with faults, recovery required, and terminated 23 Completed with faults, recovery required, and terminated 24 Running with suspended and terminated 25 Completed with suspended and terminated 26 Running with faulted, suspended, and terminated 27 Completed with faulted, suspended, and terminated 28 Running with recovery required, suspended, and terminated 29 Completed with recovery required, suspended, and terminated 30 Running with faulted, recovery required, suspended, and terminated 31 Completed with faulted, recovery required, suspended, and terminated 32 Unknown 64 - CUBE_INSTANCE States State Description 0 STATE_INITIATED 1 STATE_OPEN_RUNNING 2 STATE_OPEN_SUSPENDED 3 STATE_OPEN_FAULTED 4 STATE_CLOSED_PENDING_CANCEL 5 STATE_CLOSED_COMPLETED 6 STATE_CLOSED_FAULTED 7 STATE_CLOSED_CANCELLED 8 STATE_CLOSED_ABORTED 9 STATE_CLOSED_STALE 10 STATE_CLOSED_ROLLED_BACK DLV_MESSAGE States State Description 0 STATE_UNRESOLVED 1 STATE_RESOLVED 2 STATE_HANDLED 3 STATE_CANCELLED 4 STATE_MAX_RECOVERED Since now in 11g the Invoke_Messages table is not there so to distinguish between a new message (Invoke) and callback (DLV) and there is DLV_TYPE column that defines the type of message: DLV_TYPE States State Description 1 Invoke Message 2 DLV Message MEDIATOR_INSTANCE STATE Description  0  No faults but there still might be running instances  1  At least one case is aborted by user  2  At least one case is faulted (non-recoverable)  3  At least one case is faulted and one case is aborted  4  At least one case is in recovery required state  5 At least one case is in recovery required state and at least one is aborted  6 At least one case is in recovery required state and at least one is faulted  7 At least one case is in recovery required state, one faulted and one aborted  >=8 and < 16  Running >= 16   Stale In my next blog posting I will walk through the lifecycle of a BPEL process using the above states for the following use cases: - New BPEL process - initial Receive activity - Callback BPEL process - mid-level Receive activity As always comments and questions welcome! Deepak

    Read the article

  • Nginx and multiple wordpress instances with fastcgi under same domain

    - by damnsweet
    My site is running on apache. two instances of wordpress exist under paths /tr/ and /eng/. I want to move the setup to nginx but could not manage to get it working. My setup consists of nging 0.7.66, php 5.3.2, and php-fpm. /tr/ and /eng/ are two separate wordpress instances located under /home/istci/webapps/wordpress_tr and /home/istci/webapps/wordpress respectively. Below is the server section from nginx.conf containing only configuration for tr, yet could not get it working either. server { listen 80; server_name www.example.com; charset utf-8; location ~ ^/$ { rewrite ^(.+)$ http://www.example.com/tr/ permanent; } location ~ /tr/.*php$ { fastcgi_pass unix:/home/istci/var/run/wptr.sock; fastcgi_index index.php; fastcgi_param SCRIPT_FILENAME /home/istci/webapps/wordpress_tr$fastcgi_script_name; fastcgi_param QUERY_STRING $query_string; fastcgi_param REQUEST_METHOD $request_method; fastcgi_param CONTENT_TYPE $content_type; fastcgi_param CONTENT_LENGTH $content_length; fastcgi_param SCRIPT_NAME $fastcgi_script_name; fastcgi_param REQUEST_URI $request_uri; fastcgi_param DOCUMENT_URI $document_uri; fastcgi_param DOCUMENT_ROOT $document_root; fastcgi_param SERVER_PROTOCOL $server_protocol; fastcgi_param GATEWAY_INTERFACE CGI/1.1; fastcgi_param SERVER_SOFTWARE nginx/$nginx_version; fastcgi_param REMOTE_ADDR $remote_addr; fastcgi_param REMOTE_PORT $remote_port; fastcgi_param SERVER_ADDR $server_addr; fastcgi_param SERVER_PORT $server_port; fastcgi_param SERVER_NAME $server_name; # required if PHP was built with --enable-force-cgi-redirect fastcgi_param REDIRECT_STATUS 200; } location /tr/ { root /home/istci/webapps/wordpress_tr/; index index.php index.html index.htm; if (!-e $request_filename) { rewrite ^(.+)$ /tr/index.php?q=$1 last; break; } if (-f $request_filename) { expires 30d; break; } } } php-fpm listens on unix:/home/istci/var/run/wptr.sock. running it in debug-mode shows no active handlers, which means no connection is made to unix socket from nginx. nginx access logs: 127.0.0.1 - - [09/Jun/2010:03:45:11 -0500] "GET /tr/ HTTP/1.0" 404 20 "-" "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.2.4) Gecko/20100527 Firefox/3.6.4" nginx debug logs : 2010/06/09 03:38:53 [notice] 6922#0: built by gcc 4.1.2 20080704 (Red Hat 4.1.2-48) 2010/06/09 03:38:53 [notice] 6922#0: OS: Linux 2.6.18-164.9.1.el5PAE 2010/06/09 03:38:53 [notice] 6922#0: getrlimit(RLIMIT_NOFILE): 4096:4096 2010/06/09 03:38:53 [notice] 6923#0: start worker processes 2010/06/09 03:38:53 [notice] 6923#0: start worker process 6924 2010/06/09 03:38:53 [notice] 6923#0: start worker process 6925 2010/06/09 03:39:01 [notice] 6925#0: *1 "^(.+)$" matches "/tr/", client: 127.0.0.1, server: www.example.com, request: "GET /tr/ HTTP/1.0", host: "www.example.com" 2010/06/09 03:39:01 [notice] 6925#0: *1 rewritten data: "/tr/index.php", args: "q=/tr/", client: 127.0.0.1, server: www.example.com, request: "GET /tr/ HTTP/1.0", host: "www.example.com" Any clues about what is wrong with my configuration? Thanks.

    Read the article

  • Most efficient way to update attribute of one instance

    - by Begbie00
    Hi all - I'm creating an arbitrary number of instances (using for loops and ranges). At some event in the future, I need to change an attribute for only one of the instances. What's the best way to do this? Right now, I'm doing the following: 1) Manage the instances in a list. 2) Iterate through the list to find a key value. 3) Once I find the right object within the list (i.e. key value = value I'm looking for), change whatever attribute I need to change. for Instance within ListofInstances: if Instance.KeyValue == SearchValue: Instance.AttributeToChange = 10 This feels really inefficient: I'm basically iterating over the entire list of instances, even through I only need to change an attribute in one of them. Should I be storing the Instance references in a structure more suitable for random access (e.g. dictionary with KeyValue as the dictionary key?) Is a dictionary any more efficient in this case? Should I be using something else? Thanks, Mike

    Read the article

  • List of all states from COMPOSITE_INSTANCE, CUBE_INSTANCE, DLV_MESSAGE tables

    - by Deepak Arora
    In many of my engagements I get asked repeatedly about the states of the composites in 11g and how to decipher them, especially when we are troubleshooting issues around purging. I have compiled a list of all the states from the COMPOSITE_INSTANCE, CUBE_INSTANCE, DLV_MESSAGE and MEDIATOR_INSTANCE tables. These are the primary tables that are used when using BPEL composites and how they are used with the ECID.  Composite State Values COMPOSITE_INSTANCE States State Description 0 Running 1 Completed 2 Running with faults 3 Completed with faults 4 Running with recovery required 5 Completed with recovery required 6 Running with faults and recovery required 7 Completed with faults and recovery required 8 Running with suspended 9 Completed with suspended 10 Running with faults and suspended 11 Completed with faults and suspended 12 Running with recovery required and suspended 13 Completed with recovery required and suspended 14 Running with faults, recovery required, and suspended 15 Completed with faults, recovery required, and suspended 16 Running with terminated 17 Completed with terminated 18 Running with faults and terminated 19 Completed with faults and terminated 20 Running with recovery required and terminated 21 Completed with recovery required and terminated 22 Running with faults, recovery required, and terminated 23 Completed with faults, recovery required, and terminated 24 Running with suspended and terminated 25 Completed with suspended and terminated 26 Running with faulted, suspended, and terminated 27 Completed with faulted, suspended, and terminated 28 Running with recovery required, suspended, and terminated 29 Completed with recovery required, suspended, and terminated 30 Running with faulted, recovery required, suspended, and terminated 31 Completed with faulted, recovery required, suspended, and terminated 32 Unknown 64 - Normal 0 false false false EN-CA X-NONE X-NONE MicrosoftInternetExplorer4 /* Style Definitions */ table.MsoNormalTable {mso-style-name:"Table Normal"; mso-tstyle-rowband-size:0; mso-tstyle-colband-size:0; mso-style-noshow:yes; mso-style-priority:99; mso-style-qformat:yes; mso-style-parent:""; mso-padding-alt:0in 5.4pt 0in 5.4pt; mso-para-margin:0in; mso-para-margin-bottom:.0001pt; mso-pagination:widow-orphan; font-size:11.0pt; font-family:"Calibri","sans-serif"; mso-ascii-font-family:Calibri; mso-ascii-theme-font:minor-latin; mso-fareast-font-family:"Times New Roman"; mso-fareast-theme-font:minor-fareast; mso-hansi-font-family:Calibri; mso-hansi-theme-font:minor-latin; mso-bidi-font-family:"Times New Roman"; mso-bidi-theme-font:minor-bidi;} Any value in the range of 32 to 63 indicates that the composite instance state has not been enabled, but the instance state is updated for faults, aborts, etc. CUBE_INSTANCE States State Description 0 STATE_INITIATED 1 STATE_OPEN_RUNNING 2 STATE_OPEN_SUSPENDED 3 STATE_OPEN_FAULTED 4 STATE_CLOSED_PENDING_CANCEL 5 STATE_CLOSED_COMPLETED 6 STATE_CLOSED_FAULTED 7 STATE_CLOSED_CANCELLED 8 STATE_CLOSED_ABORTED 9 STATE_CLOSED_STALE 10 STATE_CLOSED_ROLLED_BACK DLV_MESSAGE States State Description 0 STATE_UNRESOLVED 1 STATE_RESOLVED 2 STATE_HANDLED 3 STATE_CANCELLED 4 STATE_MAX_RECOVERED Since now in 11g the Invoke_Messages table is not there so to distinguish between a new message (Invoke) and callback (DLV) and there is DLV_TYPE column that defines the type of message: DLV_TYPE States State Description 1 Invoke Message 2 DLV Message MEDIATOR_INSTANCE STATE Description  0  No faults but there still might be running instances  1  At least one case is aborted by user  2  At least one case is faulted (non-recoverable)  3  At least one case is faulted and one case is aborted  4  At least one case is in recovery required state  5 At least one case is in recovery required state and at least one is aborted  6 At least one case is in recovery required state and at least one is faulted  7 At least one case is in recovery required state, one faulted and one aborted  >=8 and < 16  Running >= 16   Stale In my next blog posting I will walk through the lifecycle of a BPEL process using the above states for the following use cases: - New BPEL process - initial Receive activity - Callback BPEL process - mid-level Receive activity As always comments and questions welcome! Deepak

    Read the article

  • How to open a pdf or a djvu on an specific instance of a document viewer?

    - by ciro
    I want using any free Linux document viewer that does both pdf and djvu (Okular, Evince, etc.) to do the following: magic_command('document-viewer','./11.pdf','instance1') magic_command('document-viewer','./21.djvu','instance2') two instances (windows) of document-viewer are opened one with 11.pdf and the other with 21.djvu then: magic_command('document-viewer','./12.djvu','instance1') magic_command('document-viewer','./22.pdf','instance2') the first instance (window) of document-viewer loads 12.djvu the second instance (window) of document-viewer loads 22.pdf

    Read the article

  • Canonical configuration for multiple-instance Postfix?

    - by threecheeseopera
    I recently attempted to reconfigure an existing single-instance Postfix server (multi-homed) to support multiple MTA instances, but failed miserably (read: open relay, nastygram from ISP). It appears that there are several methods that can be used to accomplish this, and the various (and numerous) secondary sources that I found online were 'all over the map' with respect to which they used/glued together. Can anyone provide a working configuration (or tips) that use postmulti to manage a multi-instance Postfix setup?

    Read the article

  • Unable to terminate extra EC2 instances

    - by Deborah Cole
    I'm just setting up my AWS server & I'm trying to use the EC2 Console to terminate some extra instances that I generated via the AWS for Eclipse toolkit's New Project AWS Java Web Project utility. Unfortunately, every time I stop, then terminate such an instance via the EC2 Console, it automatically recreates & reactivates itself! I really don't want to be paying for 4 dev systems when I only need 1, so can somebody please clue me in? Please explain gently... I'm new to this environment.

    Read the article

  • Apache, and logrotate instances

    - by OlivierDofus
    Hi! If I have one website and I want to rotate its logs, there's one instance of logrotate that is launched. There are as many logrotate instances launched as they are virtual websites. Here you can find mod_log_rotate, with a 1.3 version and (only) 2.0 version: http://www.hexten.net/wiki/index.php/Mod-log-rotate It's 6 years old. Is there something new, or maybe is there something like that in recent Apache versions? I didn't find anything like that, and I don't know if this code is still "usable" for recent Apache versions (2.2.x) Don't hesitate to edit my post to make it proper English, thanks a lot!

    Read the article

  • Automatically Snapshoting AWS instances (or other back up strategy)

    - by user1172468
    I just realized that my aws instance count has risen into the double digits. I'm currently backing portions of my folders and dbs and moving them off to a backup instance. What I think I should be doing is taking a snapshot of the instances (automatically) and persisting them on S3 so I have a running 7 day collection of daily backups. There is a question asking the same thing here, however the answers don't go into depth. So the closest answer seems to be: use a cron job to snapshot the instance. So do I run the cron job on the instance itself? or do I have a micro instance to run these snapshots? Could I get an example script or the command for say a linux flavor? what software must I have installed to get this to run? Thanks.

    Read the article

  • Text Editor that hilights all instances of selection for Mac

    - by jedierikb
    On Windows, I use Notepad++ which has the great feature of when I select a word, all instances of that word are also highlighted in the same document. I have found it very helpful for finding patterns in giant log files. I am wondering if there is a similar feature in a text editor on the Mac. I have looked into the documentation for TextWrangler and TextMate to no avail. Hopefully there is a way to do this so I can be more productive when working on a Mac. -- Note: in Notepad++ you do not have to do a keyboard shortcut to make this work... you just select some text and it does the highlighting for you automatically.

    Read the article

  • 2 Virtual instances for RSA and X509 authentication on IIS web server

    - by VKumar
    I am new to this concept. Actually, we have RSA SecureID login and normal login for our application. Recently, i have implemented X509 authentication. Now, the requirement is to create 2 instances for RSA access and X509 authentication on the same IIS webserver. If X509 authentication fails, the user should be redirected to RSA login page which should be on the same Webserver. Can you please help me out with some ideas? Thank you in advance.

    Read the article

  • Two Ubuntu Instances (Guest) networking with XP (Host) in virtual box v3.1.4

    - by EnthuCrazy
    So here is my current objective: I need to create two guest instances of Ubuntu Desktop 9.10 in VirtualBox on a WindowsXP host. (This is needed for communications later on.) (this step is almost done) I need to establish networking between all three OS's, the host and two guests (Guest1 - Host - Guest2). I know that generally, to establish networking between Windows host and Ubuntu guest, we set up a bridge connection. But here there are two guests, and primarily I need networking between the two guests (Ubuntu to Ubuntu). So will there need to be a change in tap0 and tap1 interfaces when we set up a bridge, or is there a better way to implement this? Please explain the procedure.

    Read the article

  • AWS Elastic load balancer doesn't decrease instances from Alarm Trigger

    - by jchysk
    I have a load balancer that I created an auto-scaling-group and launch-config for. I created the auto-scaling-group with a min-size of 1 and max size of 20. I have a scaledown policy: as-put-scaling-policy SBMScaleDownPolicy --auto-scaling-group SBMAutoScaleGroup --adjustment=-1 --type ChangeInCapacity --cooldown 300 Then I set up an alarm: mon-put-metric-alarm SBMLowCPUAlarm --comparison-operator LessThanThreshold --evaluation-periods 1 --metric-name CPUUtilization --namespace "AWS/EC2" --period 600 --statistic Average --threshold 35 --alarm-actions arn:aws:autoscaling:us-east-1:policystuffhere:autoScalingGroupName/SBMAutoScaleGroup:policyName/SBMScaleDownPolicy --dimensions "AutoScalingGroupName=SBMAutoScaleGroup" When average CPU usage over 10 minutes is under 35, in CloudFront the alarm shows up as "In Alarm State" but doesn't decrease the number of instances. Also, if there's only one instance running it'll spin up another to 2 even if a scale up alarm isn't hit. It seems like the default value is just set to 2 somehow. How can I change this?

    Read the article

  • Schedule EC2 instances

    - by mattcodes
    I want to be able to schedule some simple EC2 EBS backed instances (already configured) to start at 8am and stop at 4pm. This is only time I'll be using my integration server. Is there a simple services (paid or not) that I can use to handle this. All I found so far is to buy a cheap VPS at linode or somewhere and install ec2 tools and schedule via crontab, but what a PITA that is to. On the other end is something enterprisey like Rightscale but not my idea of simple.

    Read the article

  • Deploy multiple django instances on one Host [migrated]

    - by tvn
    I am trying to setup multiple Django instances on one Host with lighttpd. My problem is to get Djangos FCGI working on subdirectories served by my Webserver. So my aim is the following: www.myhost.org/django0 - django1.fcgi on localhost:3000 www.myhost.org/django1 - django2.fcgi on localhost:3001 www.myhost.org/django2 - django3.fcgi on localhost:3002 Unfortunately the following configuration doesn't even work for one: $HTTP["url"] =~ "^/django0/static($|/)" { server.document-root = "/home/django0/django/static/" } $HTTP["url"] =~ "^/django0/media($|/)" { server.document-root = "/usr/lib/python2.7/dist-packages/django/contrib/admin/media/" } $HTTP["url"] =~ "^/django0($|/)" { proxy.server = ( "" => ( ( "host" => "127.0.0.1", "port" => "3001", "check-local" => "disable", ) ) ) } The only response I get is an 404 and even this takes a long time till I get this. I found nothing suspicious neither in the access.log nor in the error.log.

    Read the article

  • SVN - managing .htaccess file on various instances

    - by user178087
    I have a question and need a suggestion. We have to manage .htaccess file on three different instances - dev, qa and prod. Currently we have tortoise configured for scm. The issue we are facing is that .htaccess for dev & qa is different from the prod. At present, we have to manually merge differences from dev .htaccess to prod .htaccess. Is there any alternative way of managing this file without this manual process since it is error prone. Any suggestions in this regard will be highly appreciated

    Read the article

  • Consolidate information held in a number of SQL Server Express Instances

    - by user321271
    Hi, I'm trying to determine the best architecture for creating an oData web service for information held in a number of SQL Server Express instances. The web service should provide a consolidated view of the data. All the SQL Server Express instances have the same DB schema. I was initially planning to use SQL server replication however as I understand it, SQL Server 2008 Express cannot be used as a publisher. Any help or suggestions would be appreciated.

    Read the article

  • Following the Thread in OSB

    - by Antony Reynolds
    Threading in OSB The Scenario I recently led an OSB POC where we needed to get high throughput from an OSB pipeline that had the following logic: 1. Receive Request 2. Send Request to External System 3. If Response has a particular value   3.1 Modify Request   3.2 Resend Request to External System 4. Send Response back to Requestor All looks very straightforward and no nasty wrinkles along the way.  The flow was implemented in OSB as follows (see diagram for more details): Proxy Service to Receive Request and Send Response Request Pipeline   Copies Original Request for use in step 3 Route Node   Sends Request to External System exposed as a Business Service Response Pipeline   Checks Response to Check If Request Needs to Be Resubmitted Modify Request Callout to External System (same Business Service as Route Node) The Proxy and the Business Service were each assigned their own Work Manager, effectively giving each of them their own thread pool. The Surprise Imagine our surprise when, on stressing the system we saw it lock up, with large numbers of blocked threads.  The reason for the lock up is due to some subtleties in the OSB thread model which is the topic of this post.   Basic Thread Model OSB goes to great lengths to avoid holding on to threads.  Lets start by looking at how how OSB deals with a simple request/response routing to a business service in a route node. Most Business Services are implemented by OSB in two parts.  The first part uses the request thread to send the request to the target.  In the diagram this is represented by the thread T1.  After sending the request to the target (the Business Service in our diagram) the request thread is released back to whatever pool it came from.  A multiplexor (muxer) is used to wait for the response.  When the response is received the muxer hands off the response to a new thread that is used to execute the response pipeline, this is represented in the diagram by T2. OSB allows you to assign different Work Managers and hence different thread pools to each Proxy Service and Business Service.  In out example we have the “Proxy Service Work Manager” assigned to the Proxy Service and the “Business Service Work Manager” assigned to the Business Service.  Note that the Business Service Work Manager is only used to assign the thread to process the response, it is never used to process the request. This architecture means that while waiting for a response from a business service there are no threads in use, which makes for better scalability in terms of thread usage. First Wrinkle Note that if the Proxy and the Business Service both use the same Work Manager then there is potential for starvation.  For example: Request Pipeline makes a blocking callout, say to perform a database read. Business Service response tries to allocate a thread from thread pool but all threads are blocked in the database read. New requests arrive and contend with responses arriving for the available threads. Similar problems can occur if the response pipeline blocks for some reason, maybe a database update for example. Solution The solution to this is to make sure that the Proxy and Business Service use different Work Managers so that they do not contend with each other for threads. Do Nothing Route Thread Model So what happens if there is no route node?  In this case OSB just echoes the Request message as a Response message, but what happens to the threads?  OSB still uses a separate thread for the response, but in this case the Work Manager used is the Default Work Manager. So this is really a special case of the Basic Thread Model discussed above, except that the response pipeline will always execute on the Default Work Manager.   Proxy Chaining Thread Model So what happens when the route node is actually calling a Proxy Service rather than a Business Service, does the second Proxy Service use its own Thread or does it re-use the thread of the original Request Pipeline? Well as you can see from the diagram when a route node calls another proxy service then the original Work Manager is used for both request pipelines.  Similarly the response pipeline uses the Work Manager associated with the ultimate Business Service invoked via a Route Node.  This actually fits in with the earlier description I gave about Business Services and by extension Route Nodes they “… uses the request thread to send the request to the target”. Call Out Threading Model So what happens when you make a Service Callout to a Business Service from within a pipeline.  The documentation says that “The pipeline processor will block the thread until the response arrives asynchronously” when using a Service Callout.  What this means is that the target Business Service is called using the pipeline thread but the response is also handled by the pipeline thread.  This implies that the pipeline thread blocks waiting for a response.  It is the handling of this response that behaves in an unexpected way. When a Business Service is called via a Service Callout, the calling thread is suspended after sending the request, but unlike the Route Node case the thread is not released, it waits for the response.  The muxer uses the Business Service Work Manager to allocate a thread to process the response, but in this case processing the response means getting the response and notifying the blocked pipeline thread that the response is available.  The original pipeline thread can then continue to process the response. Second Wrinkle This leads to an unfortunate wrinkle.  If the Business Service is using the same Work Manager as the Pipeline then it is possible for starvation or a deadlock to occur.  The scenario is as follows: Pipeline makes a Callout and the thread is suspended but still allocated Multiple Pipeline instances using the same Work Manager are in this state (common for a system under load) Response comes back but all Work Manager threads are allocated to blocked pipelines. Response cannot be processed and so pipeline threads never unblock – deadlock! Solution The solution to this is to make sure that any Business Services used by a Callout in a pipeline use a different Work Manager to the pipeline itself. The Solution to My Problem Looking back at my original workflow we see that the same Business Service is called twice, once in a Routing Node and once in a Response Pipeline Callout.  This was what was causing my problem because the response pipeline was using the Business Service Work Manager, but the Service Callout wanted to use the same Work Manager to handle the responses and so eventually my Response Pipeline hogged all the available threads so no responses could be processed. The solution was to create a second Business Service pointing to the same location as the original Business Service, the only difference was to assign a different Work Manager to this Business Service.  This ensured that when the Service Callout completed there were always threads available to process the response because the response processing from the Service Callout had its own dedicated Work Manager. Summary Request Pipeline Executes on Proxy Work Manager (WM) Thread so limited by setting of that WM.  If no WM specified then uses WLS default WM. Route Node Request sent using Proxy WM Thread Proxy WM Thread is released before getting response Muxer is used to handle response Muxer hands off response to Business Service (BS) WM Response Pipeline Executes on Routed Business Service WM Thread so limited by setting of that WM.  If no WM specified then uses WLS default WM. No Route Node (Echo functionality) Proxy WM thread released New thread from the default WM used for response pipeline Service Callout Request sent using proxy pipeline thread Proxy thread is suspended (not released) until the response comes back Notification of response handled by BS WM thread so limited by setting of that WM.  If no WM specified then uses WLS default WM. Note this is a very short lived use of the thread After notification by callout BS WM thread that thread is released and execution continues on the original pipeline thread. Route/Callout to Proxy Service Request Pipeline of callee executes on requestor thread Response Pipeline of caller executes on response thread of requested proxy Throttling Request message may be queued if limit reached. Requesting thread is released (route node) or suspended (callout) So what this means is that you may get deadlocks caused by thread starvation if you use the same thread pool for the business service in a route node and the business service in a callout from the response pipeline because the callout will need a notification thread from the same thread pool as the response pipeline.  This was the problem we were having. You get a similar problem if you use the same work manager for the proxy request pipeline and a business service callout from that request pipeline. It also means you may want to have different work managers for the proxy and business service in the route node. Basically you need to think carefully about how threading impacts your proxy services. References Thanks to Jay Kasi, Gerald Nunn and Deb Ayers for helping to explain this to me.  Any errors are my own and not theirs.  Also thanks to my colleagues Milind Pandit and Prasad Bopardikar who travelled this road with me. OSB Thread Model Great Blog Post on Thread Usage in OSB

    Read the article

  • How to run multiple instances of Tor?

    - by Ed
    I'm trying to set up a special proxy server (running Windows). It will have several instances of Privoxy and Tor running and my app will choose which Privoxy instance to send HTTP requests to depending on the load. Privoxy will then forward them to Tor. I'm using srvany.exe to create the services. At the moment I'm running 3 Privoxy and 3 Tor services (I copied the binaries to different folders). Each Privoxy service is listening to its own port (8118, 8119, 8120). I can see them listening in a port scanner. This is the application path (for srvany in registry) for the 1st service: C:\Anonymiser\Privoxy 01\privoxy.exe --service I've also configured the Tor services to listen to different ports (9050, 9052, 9054). This is the application path for the 1st service: C:\Anonymiser\Tor 01\tor.exe -f "C:\Anonymiser\Tor 01\torrc" The problem is, when I start the Tor services, only the first service I start is listening to its port. The others aren't listening. They listen if I run them separately. Any ideas what could be wrong? How can I make all 3 services listen on their assigned ports? This is one of my Privoxy configs: confdir . logdir . logfile privoxy.log debug 1 # show each GET/POST/CONNECT request debug 4096 # Startup banner and warnings debug 8192 # Errors - we highly recommended enabling this listen-address localhost:8118 toggle 0 enable-remote-toggle 0 enable-remote-http-toggle 0 enable-edit-actions 1 buffer-limit 4096 forwarded-connect-retries 0 forward-socks4a / localhost:9050 . This is one of my Tor configs: ControlPort 9051 Log notice stdout SocksListenAddress localhost SocksPort 9050 EDIT: Found a workaround. The Tor binary wants a lock on a file in the AppData folder. Because all of them want a lock on the same file, only the first one I start will be working. The workaround is to run each Tor instance under a different account. Not the best solution, but it works.

    Read the article

  • Domino nchronos.exe multiple instances causing server to die, and Sametime problems

    - by Kevin
    I've had this problem for a few months now. I thought it started when I installed the Traveller software on the server to add ActiveSync support, but I removed that and the problem still persists. Basically new instances of "nchronos.exe" keeps spawning (and not ending), so over a period of a few days the server eventually gets drowned in nchronos.exe processes, stops responding and I need to kill Domino. My process count the last time was up at about 330, and when I killed it and restarted the Domino my process count went to 160. I'm running Domino 8.5.1 with Fix Pack 2. I don't know if it's relevant, but my Domino server was also acting as a Sametime server. At around the same time that nchronos started playing up sametime also stopped working. None of my users can connect to sametime and in the domino log it keeps telling me "stpolicy.exe" has terminated. I've googled for that and tried a few things, but nothing seems to make sametime work again. Any thoughts?? Cheers, Kevin

    Read the article

  • Domino nchronos.exe multiple instances causing server to die, and Sametime problems

    - by Kevin
    I've had this problem for a few months now. I thought it started when I installed the Traveller software on the server to add ActiveSync support, but I removed that and the problem still persists. Basically new instances of "nchronos.exe" keeps spawning (and not ending), so over a period of a few days the server eventually gets drowned in nchronos.exe processes, stops responding and I need to kill Domino. My process count the last time was up at about 330, and when I killed it and restarted the Domino my process count went to 160. I'm running Domino 8.5.1 with Fix Pack 2. I don't know if it's relevant, but my Domino server was also acting as a Sametime server. At around the same time that nchronos started playing up sametime also stopped working. None of my users can connect to sametime and in the domino log it keeps telling me "stpolicy.exe" has terminated. I've googled for that and tried a few things, but nothing seems to make sametime work again. Any thoughts?? Cheers, Kevin

    Read the article

  • Production monitoring for EC2 instances

    - by Janine
    I'm setting up my first production instance on EC2 and want to make sure I have all necessary monitoring in place. There are three different types of things I want to monitor: Is the instance running? EC2 instances can be terminated without warning if the underlying hardware fails, and as far as I know they aren't automatically restarted. So if not, start it back up. Is UNIX running properly? This is the usual stuff about CPU load, disk space, etc. Is the website responding? If not, restart it. I initially set up Nagios on a physical server outside the cloud, but it is really only helpful for item 2. It can tell me if the instance is gone or if the website is not responding, but as far as I can tell it can't execute any commands to fix the situation. My Googling on this subject has yielded a plethora of options - Cacti, Monit, God, Ganglia, and probably more I'm forgetting now. I don't have time to research them all. I am aware of Amazon's Cloudwatch but it doesn't seem to do anything that my Nagios installation doesn't already do. If you already have something like this in place, can you please share what has worked well for you?

    Read the article

  • Amazon EC2: Instances, IPs and a wordpress blog (LAMP)

    - by JustinXXVII
    I had a link to my blog posted on Reddit yesterday and MySQL crashed on my EC2 Micro instance. I know I didn't have that many visitors because I used a marketing link that tracks hits. The link got 167 hits over the course of the last 18 hours, and MySQL crashed twice. So anyway, 167 visits is not a lot, so I've done some short term optimizations like restricting the number of Apache threads to limit the MySQL calls. I also set up WP Super Cache to serve static content. Soon I'm going to offload all of my images to S3 or CloudFront. So this leads me to my question. If this doesn't seem to help, and if i have another traffic "spike", how do AMIs work when you have a MySQL database? I think I understand that if you have more than one instance and assign the same Elastic IP to both of them, the incoming traffic gets distributed among both. But what happens when the MySQL database gets updated on one of the instances? I just need to wrap my mind around what happens when I create an AMI and then launch a new instance to help with traffic. Thanks for your suggestions.

    Read the article

< Previous Page | 17 18 19 20 21 22 23 24 25 26 27 28  | Next Page >