Search Results

Search found 30270 results on 1211 pages for 'bart read'.

Page 156/1211 | < Previous Page | 152 153 154 155 156 157 158 159 160 161 162 163  | Next Page >

  • Mystery in Ruby sinatra

    - by JVK
    I have the following Sinatra code: post '/bucket' do # determine if this call is coming from filling out web form is_html = request.content_type.to_s.downcase.eql?('application/x-www-form-urlencoded') # If this is a curl call, then get the params differently unless is_html params = JSON.parse(request.env["rack.input"].read) end p params[:name] end If I call this using Curl, params has values, but when this is called via a web form, then params is nil and params[:name] has nothing. I spent several hours figuring out why it happens and asked help from other people, but no one could really find out what is going on. One thing to note is, if I comment out this line: params = JSON.parse(request.env["rack.input"].read) then params has the correct value for "web-form" posting. Actually, the goal is to get the params value if this code is being called by CURL call, so I used: params = JSON.parse(request.env["rack.input"].read) but it messed up the web-form posting. Can anyone solve this mystery?

    Read the article

  • Problems with 'while' loop and 'for' loop when reading through file

    - by David Beckham
    I wasted plenty of hours trying to figure out the problem but no luck. Tried asking the TA at my school, but he was useless. I am a beginner and I know there are a lot of mistakes in it, so it would be great if I can get some detail explanation as well. Anyways, basically what I am trying to do with the following function is: Use while loop to check and see if random_string is in TEXT, if not return NoneType if yes, then use a for loop to read lines from that TEXT and put it in list, l1. then, write an if statement to see if random_string is in l1. if it is, then do some calculations. else read the next line Finally, return the calculations as a whole. TEXT = open('randomfile.txt') def random (TEXT, random_string): while random_string in TEXT: for lines in TEXT: l1=TEXT.readline().rsplit() if random_string in l1: ''' do some calculations ''' else: TEXT.readline() #read next line??? return #calculations return None

    Read the article

  • Need help making this code more efficient

    - by Rendicahya
    I always use this method to easily read the content of a file. Is it efficient enough? Is 1024 good for the buffer size? public static String read(File file) { FileInputStream stream = null; StringBuilder str = new StringBuilder(); try { stream = new FileInputStream(file); } catch (FileNotFoundException e) { } FileChannel channel = stream.getChannel(); ByteBuffer buffer = ByteBuffer.allocate(1024); try { while (channel.read(buffer) != -1) { buffer.flip(); while (buffer.hasRemaining()) { str.append((char) buffer.get()); } buffer.rewind(); } } catch (IOException e) { } finally { try { channel.close(); stream.close(); } catch (IOException e) { } } return str.toString(); }

    Read the article

  • F# match char values

    - by rwallace
    I'm trying to match an integer expression against character literals, and the compiler complains about type mismatch. let rec read file includepath = let ch = ref 0 let token = ref 0 use stream = File.OpenText file let readch() = ch := stream.Read() let lex() = match !ch with | '!' -> readch() | _ -> token := !ch ch has to be an int because that's what stream.Read returns in order to use -1 as end of file marker. If I replace '!' with int '!' it still doesn't work. What's the best way to do this?

    Read the article

  • Lua not producing table of functions (IO API)

    - by ArtOfCode
    I'm working on a basic project in Lua. I've been trying to get data from files using the IO API (as defined here), but when I open a file and give it a handle, it doesn't seem to return a table of functions. The (erroneous bit of) code: local unread = fs.list("email/"..from.."/") local send = "" for _,file in ipairs(unread) do local handle = io.open(file,"r") local text = handle:read("*a") send = send .. text .. "\n" handle.close() fs.delete(file) end The fs you see on the first line is a professional filesystem wrapper round the IO API, not my work and perfectly error-free, so that's not the problem. However, when I try to read the file (handle:read()), it throws "attempt to index nil". Tracing it, it turns out that handle itself is nil. Any ideas?

    Read the article

  • using different string files in android

    - by boreas
    I'm porting my iPhone app to android and I'm having a problem with the string files now. The app is a translation tool and users can switch the languages, so all the localized strings are in both languages and they are independent from what locale the OS is running. For iOS version I have different files like de.strings, en.strings and fr.strings and so on. For every target with specified language pair I read the strings from the string tables, e.g. for de-fr I will include de.strings and fr.strings in project and set the name of the string tables in the info-list file and read strings from them. In the end I have one project containing different targets (with different info-list files) and all are well configured. I'm intending to do the same on android platform, but Is only one strings.xml allowed per project? How do I set different build target? How do I specify per target which strings.xml it should read?

    Read the article

  • Low-overhead way to access the memory space of a traced process?

    - by vovick
    Hello all. I'm looking for an efficient way to access(for both read and write operations) the memory space of my ptraced child process. The size of blocks being accessed may vary from several bytes up to several megabytes in size, so using the ptrace call with PTRACE_PEEKDATA and PTRACE_POKEDATA which read only one word at a time and switch context every time they're called seems like a pointless waste of resources. The only one alternative solution I could find, though, was the /proc/<pid>/mem file, but it has long since been made read only. Is there any other (relatively simple) way to do that job? The ideal solution would be to somehow share the address space of my child process with its parent and then use the simple memcpy call to copy data I need in both directions, but I have no clues how to do it and where to begin. Any ideas?

    Read the article

  • Reading function pointer syntax

    - by bobobobo
    Everytime I look at a C function pointer, my eyes glaze over. I can't read them. From here, here are 2 examples of function pointer TYPEDEFS: typedef int (*AddFunc)(int,int); typedef void (*FunctionFunc)(); Now I'm used to something like: typedef vector<int> VectorOfInts ; Which I read as typedef vector<int> /* as */ VectorOfInts ; But I can't read the above 2 typedefs. The bracketing and the asterisk placement, it's just not logical. Why is the * beside the word AddFunc..?

    Read the article

  • Get result type of function

    - by Robert
    I want to specialize a template function declared as: template<typename Type> Type read(std::istream& is); I then have a lot of static implementations static int read_integer(std::istream& is); a.s.o. Now I'd like to do a macro so that specialization of read is as simple as: SPECIALIZE_READ(read_integer) So I figured I'd go the boost::function_traits way and declare SPECIALIZE_READ as: #define SPECIALIZE_READ(read_function) \ template<> boost::function_traits<read_function>::result_type read(std::istream& is) { \ return read_function(is); \ } but VC++ (2008) compiler complains with: 'boost::function_traits' : 'read_integer' is not a valid template type argument for parameter 'Function' Ideas ?

    Read the article

  • Optimal strategy to make a C++ hash table, thread safe

    - by Ajeet
    (I am interested in design of implementation NOT a readymade construct that will do it all.) Suppose we have a class HashTable (not hash-map implemented as a tree but hash-table) and say there are eight threads. Suppose read to write ratio is about 100:1 or even better 1000:1. Case A) Only one thread is a writer and others including writer can read from HashTable(they may simply iterate over entire hash table) Case B) All threads are identical and all could read/write. Can someone suggest best strategy to make the class thread safe with following consideration 1. Top priority to least lock contention 2. Second priority to least number of locks My understanding so far is thus : One BIG reader-writer lock(semaphore). Specialize the semaphore so that there could be eight instances writer-resource for case B, where each each writer resource locks one row(or range for that matter). (so i guess 1+8 mutexes) Please let me know if I am thinking on the correct line, and how could we improve on this solution.

    Read the article

  • function declaration

    - by robUK
    Hello, gcc 4.1.2 c89 I am reviewing some code and I have come across the following function. I have never seen a function declared like this before. There are no data types for the paraemeters. My best guess is that the function is using a list of data types separated by semi-colons. The return type seems to be returning a function with those parameters. However, the read function is not defined anywhere. What is the advantage and purpose of declaring a function like this? Many thanks for any advice, int my_read(fd, ptr, cnt) int fd; char *ptr; unsigned cnt; { printf("Read\n"); return(read(fd, ptr, cnt)); }

    Read the article

  • Trimming strings in Go

    - by user1263980
    I'm trying to read an entire line from the console (including whitespace), then process it. Using bufio.ReadString, the newline character is read together with the input, so I came up with the following code to trim the newline character: input,_:=src.ReadString('\n') inputFmt:=input[0:len(input)-2]+"" Is there a more idiomatic way to do this? That is, is there already a library that takes care of the ending null byte when extracting substrings for you? (Yes, I know there is already a way to read a line without the newline character in go readline -> string but I'm looking more for elegant string manipulation.)

    Read the article

  • How can I set a default value when incorrect/invalid input is entered in Unix?

    - by user2889968
    i want to set the value of inputLineNumber to 20. I tried checking if no value is given by user by [[-z "$inputLineNumber"]] and then setting the value by inputLineNumber=20. The code seems to be working correct when no value is given but I also receive ./t.sh: [-z: not found as message on the console. Is there some way to not print this message. Here's my full script as well. #!/bin/sh cat /dev/null>copy.txt echo "Please enter the sentence you want to search:" read "inputVar" echo "Please enter the name of the file in which you want to search:" read "inputFileName" echo "Please enter the number of lines you want to copy:" read "inputLineNumber" [[-z "$inputLineNumber"]] || inputLineNumber=20 cat /dev/null > copy.txt for N in `grep -n $inputVar $inputFileName | cut -d ":" -f1` do LIMIT=`expr $N + $inputLineNumber` sed -n $N,${LIMIT}p $inputFileName >> copy.txt echo "-----------------------" >> copy.txt done cat copy.txt

    Read the article

  • ASP.NET MVC 3 RTM Released

    - by shiju
     The ASP.NET team has released RTM version of ASP.NET MVC 3. You can download the ASP.NET MVC 3 RTM from here and source code of ASP.NET MVC 3 can download from here. Microsoft has released the following products along with ASP.NET MVC 3.NuGetIIS Express 7.5SQL Server Compact Edition 4Web Deploy and Web Farm Framework 2.0Orchard 1.0WebMatrix 1.0 You can read more details from ScottGu's blog post Announcing release of ASP.NET MVC 3, IIS Express, SQL CE 4, Web Farm Framework, Orchard, WebMatrix .You can upgrade your ASP.NET MVC 2 projects to ASP.NET MVC 3 using MVC 3 Project Upgrade Tool. You can read more details about the MVC 3 Upgrade Tool from here. Demo Web App using ASP.NET MVC 3 RTM  You can download a demo web app using ASP.NET MVC 3 RTM from here. The demo app is explained in the below blog postsDeveloping web apps using ASP.NET MVC 3, Razor and EF Code First - Part 1Developing web apps using ASP.NET MVC 3, Razor and EF Code First - Part 2

    Read the article

  • SQL SERVER – Shrinking NDF and MDF Files – Readers’ Opinion

    - by pinaldave
    Previously, I had written a blog post about SQL SERVER – Shrinking NDF and MDF Files – A Safe Operation. After that, I have written the following blog post that talks about the advantage and disadvantage of Shrinking and why one should not be Shrinking a file SQL SERVER – SHRINKFILE and TRUNCATE Log File in SQL Server 2008. On this subject, SQL Server Expert Imran Mohammed left an excellent comment. I just feel that his comment is worth a big article itself. For everybody to read his wonderful explanation, I am posting this blog post here. Thanks Imran! Shrinking Database always creates performance degradation and increases fragmentation in the database. I suggest that you keep that in mind before you start reading the following comment. If you are going to say Shrinking Database is bad and evil, here I am saying it first and loud. Now, the comment of Imran is written while keeping in mind only the process showing how the Shrinking Database Operation works. Imran has already explained his understanding and requests further explanation. I have removed the Best Practices section from Imran’s comments, as there are a few corrections. Comments from Imran - Before I explain to you the concept of Shrink Database, let us understand the concept of Database Files. When we create a new database inside the SQL Server, it is typical that SQl Server creates two physical files in the Operating System: one with .MDF Extension, and another with .LDF Extension. .MDF is called as Primary Data File. .LDF is called as Transactional Log file. If you add one or more data files to a database, the physical file that will be created in the Operating System will have an extension of .NDF, which is called as Secondary Data File; whereas, when you add one or more log files to a database, the physical file that will be created in the Operating System will have the same extension as .LDF. The questions now are, “Why does a new data file have a different extension (.NDF)?”, “Why is it called as a secondary data file?” and, “Why is .MDF file called as a primary data file?” Answers: Note: The following explanation is based on my limited knowledge of SQL Server, so experts please do comment. A data file with a .MDF extension is called a Primary Data File, and the reason behind it is that it contains Database Catalogs. Catalogs mean Meta Data. Meta Data is “Data about Data”. An example for Meta Data includes system objects that store information about other objects, except the data stored by the users. sysobjects stores information about all objects in that database. sysindexes stores information about all indexes and rows of every table in that database. syscolumns stores information about all columns that each table has in that database. sysusers stores how many users that database has. Although Meta Data stores information about other objects, it is not the transactional data that a user enters; rather, it’s a system data about the data. Because Primary Data File (.MDF) contains important information about the database, it is treated as a special file. It is given the name Primary Data file because it contains the Database Catalogs. This file is present in the Primary File Group. You can always create additional objects (Tables, indexes etc.) in the Primary data file (This file is present in the Primary File group), by mentioning that you want to create this object under the Primary File Group. Any additional data file that you add to the database will have only transactional data but no Meta Data, so that’s why it is called as the Secondary Data File. It is given the extension name .NDF so that the user can easily identify whether a specific data file is a Primary Data File or a Secondary Data File(s). There are many advantages of storing data in different files that are under different file groups. You can put your read only in the tables in one file (file group) and read-write tables in another file (file group) and take a backup of only the file group that has read the write data, so that you can avoid taking the backup of a read-only data that cannot be altered. Creating additional files in different physical hard disks also improves I/O performance. A real-time scenario where we use Files could be this one: Let’s say you have created a database called MYDB in the D-Drive which has a 50 GB space. You also have 1 Database File (.MDF) and 1 Log File on D-Drive and suppose that all of that 50 GB space has been used up and you do not have any free space left but you still want to add an additional space to the database. One easy option would be to add one more physical hard disk to the server, add new data file to MYDB database and create this new data file in a new hard disk then move some of the objects from one file to another, and put the file group under which you added new file as default File group, so that any new object that is created gets into the new files, unless specified. Now that we got a basic idea of what data files are, what type of data they store and why they are named the way they are, let’s move on to the next topic, Shrinking. First of all, I disagree with the Microsoft terminology for naming this feature as “Shrinking”. Shrinking, in regular terms, means to reduce the size of a file by means of compressing it. BUT in SQL Server, Shrinking DOES NOT mean compressing. Shrinking in SQL Server means to remove an empty space from database files and release the empty space either to the Operating System or to SQL Server. Let’s examine this through an example. Let’s say you have a database “MYDB” with a size of 50 GB that has a free space of about 20 GB, which means 30GB in the database is filled with data and the 20 GB of space is free in the database because it is not currently utilized by the SQL Server (Database); it is reserved and not yet in use. If you choose to shrink the database and to release an empty space to Operating System, and MIND YOU, you can only shrink the database size to 30 GB (in our example). You cannot shrink the database to a size less than what is filled with data. So, if you have a database that is full and has no empty space in the data file and log file (you don’t have an extra disk space to set Auto growth option ON), YOU CANNOT issue the SHRINK Database/File command, because of two reasons: There is no empty space to be released because the Shrink command does not compress the database; it only removes the empty space from the database files and there is no empty space. Remember, the Shrink command is a logged operation. When we perform the Shrink operation, this information is logged in the log file. If there is no empty space in the log file, SQL Server cannot write to the log file and you cannot shrink a database. Now answering your questions: (1) Q: What are the USEDPAGES & ESTIMATEDPAGES that appear on the Results Pane after using the DBCC SHRINKDATABASE (NorthWind, 10) ? A: According to Books Online (For SQL Server 2000): UsedPages: the number of 8-KB pages currently used by the file. EstimatedPages: the number of 8-KB pages that SQL Server estimates the file could be shrunk down to. Important Note: Before asking any question, make sure you go through Books Online or search on the Google once. The reasons for doing so have many advantages: 1. If someone else already has had this question before, chances that it is already answered are more than 50 %. 2. This reduces your waiting time for the answer. (2) Q: What is the difference between Shrinking the Database using DBCC command like the one above & shrinking it from the Enterprise Manager Console by Right-Clicking the database, going to TASKS & then selecting SHRINK Option, on a SQL Server 2000 environment? A: As far as my knowledge goes, there is no difference, both will work the same way, one advantage of using this command from query analyzer is, your console won’t be freezed. You can do perform your regular activities using Enterprise Manager. (3) Q: What is this .NDF file that is discussed above? I have never heard of it. What is it used for? Is it used by end-users, DBAs or the SERVER/SYSTEM itself? A: .NDF File is a secondary data file. You never heard of it because when database is created, SQL Server creates database by default with only 1 data file (.MDF) and 1 log file (.LDF) or however your model database has been setup, because a model database is a template used every time you create a new database using the CREATE DATABASE Command. Unless you have added an extra data file, you will not see it. This file is used by the SQL Server to store data which are saved by the users. Hope this information helps. I would like to as the experts to please comment if what I understand is not what the Microsoft guys meant. Reference: Pinal Dave (http://blog.SQLAuthority.com) Filed under: Readers Contribution, Readers Question, SQL, SQL Authority, SQL Query, SQL Scripts, SQL Server, SQL Tips and Tricks, T SQL, Technology

    Read the article

  • USB packets - receive wrong data

    - by regorianer
    i have a little python script which shows me the packets of an enocean device and does some events depending on the packet type. unfortunately it doesn't work because i'm getting wrong packets. Parts of the python script (used pySerial): Blockquote ser = serial.Serial('/dev/ttyUSB1',57600,bytesize = serial.EIGHTBITS,timeout = 1, parity = serial.PARITY_NONE , rtscts = 0) print 'clearing buffer' s = ser.read(10000) print 'start read' while 1: s = ser.read(1) for character in s: sys.stdout.write(" %s" % character.encode('hex')) print 'end' ser.close() output baudrate 57600: e0 e0 00 e0 00 e0 e0 e0 e0 e0 00 e0 e0 00 00 00 00 00 00 00 e0 e0 e0 00 00 00 00 e0 e0 e0 00 00 e0 e0 e0 e0 e0 00 e0 00 e0 e0 e0 e0 e0 00 e0 e0 00 00 00 00 00 00 e0 e0 e0 00 00 00 00 e0 e0 e0 00 00 e0 e0 e0 output baudrate 9600: a5 5a 0b 05 10 00 00 00 00 15 c4 56 20 6f a5 5a 0b 05 00 00 00 00 00 15 c4 56 20 5f linux terminal baudrate 57600: $stty -F /dev/ttyUSB1 57600 $stty < /dev/ttyUSB1 speed 57600 baud; line = 0; eof = ^A; min = 0; time = 0; -brkint -icrnl -imaxbel -opost -onlcr -isig -icanon -iexten -echo -echoe -echok -echoctl -echoke $while (true) do cat -A /dev/ttyUSB1 ; done myfile $hexdump -C myfile 00000000 4d 2d 60 4d 2d 60 5e 40 4d 2d 60 5e 40 4d 2d 60 |M-M-^@M-^@M-| 00000010 4d 2d 60 4d 2d 60 4d 2d 60 4d 2d 60 5e 40 4d 2d |M-M-M-M-^@M-| 00000020 60 4d 2d 60 5e 40 5e 40 5e 40 5e 40 5e 40 5e 40 |M-^@^@^@^@^@^@| 00000030 5e 40 4d 2d 60 4d 2d 60 4d 2d 60 5e 40 5e 40 5e |^@M-M-M-`^@^@^| 00000040 40 5e 40 4d 2d 60 4d 2d 60 4d 2d 60 |@^@M-M-M-`| 0000004c linux terminal baudrate 9600: $hexdump -C myfile2 00000000 5e 40 5e 55 4d 2d 44 56 30 4d 2d 3f 5e 40 5e 40 |^@^UM-DV0M-?^@^@| 00000010 5e 55 4d 2d 44 56 20 5f |^UM-DV _| 00000018 the specification says: 0x55 sync byte 1st 0xNNNN data length bytes (2 bytes) 0x07 opt length byte 0x01 type byte CRC, data, opt data und nochmal CRC but I'm not getting this packet structure. The output of the python script differs from the one I get via the terminal. I also wrote the python part with C, but the output is the same as with python As the USB receiver a BSC-BoR USB Receiver/Sender is used The EnOcean device is a simple button

    Read the article

  • Bob Dorr’s SQL I/O Presentation on PSS Blog

    - by Jonathan Kehayias
    In case you missed it, Bob Dorr from the PSS Team posted an amazing blog post today yesterday with all of the slides and speaker notes from his SQL Server I/O presentation.  This is a must read for and Database Professional using SQL Server. http://blogs.msdn.com/psssql/archive/2010/03/24/how-it-works-bob-dorr-s-sql-server-i-o-presentation.aspx Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!...(read more)

    Read the article

  • Bob Dorr’s SQL I/O Presentation on PSS Blog

    - by Jonathan Kehayias
    In case you missed it, Bob Dorr from the PSS Team posted an amazing blog post today yesterday with all of the slides and speaker notes from his SQL Server I/O presentation.  This is a must read for and Database Professional using SQL Server. http://blogs.msdn.com/psssql/archive/2010/03/24/how-it-works-bob-dorr-s-sql-server-i-o-presentation.aspx Share this post: email it! | bookmark it! | digg it! | reddit! | kick it! | live it!...(read more)

    Read the article

  • data source does not support server-side data paging uisng asp.net Csharp

    - by Aamir Hasan
    Yesterday some one mail me and ask about data source does not support server side data paging.So i write the the solution here please if you have got this problem read this article and see the example code this will help you a Lot.The only change you have to do is in the DataBind().Here you have used the SqlDataReader to read data retrieved from the database, but SqlDataReader is forward only. You can not traverse back and forth on it.So the solution for this is using DataAdapter and DataSet.So your function may change some what like this private void DataBind(){//for grid viewSqlCommand cmdO;string SQL = "select * from TABLE ";conn.Open();cmdO = new SqlCommand(SQL, conn);SqlDataAdapter da = new SqlDataAdapter(cmdO);DataSet ds = new DataSet();da.Fill(ds);GridView1.Visible = true;GridView1.DataSource = ds;GridView1.DataBind();ds.Dispose();da.Dispose();conn.Close();} This surely works. The reset of your code is fine. Enjoy coding.

    Read the article

  • SQLAuthority News – Don’t Be Afraid To Fool The World – Video by John Sonmez

    - by Pinal Dave
    Sometime some words and statements grabs your attention and it is hard to stop thinking about that after a while. Something similar happened a few days ago when I read the twitter statement of my friend and Pluralsight author John Sonmez. He twitted few days ago very interesting statement. “I don’t know a single successful person, who doesn’t deep down think that have the world fooled. #fooltheworld” by John Sonmez. When I read it, I was extremely intrigued by this statement. I read it many times, I shared with my family and I just could not stop interpreting this statement. It was indeed fun to read it again and again and there are so many different meanings one can take away from the statement. I know John very well, he is a  wonderful person and have very positive energy for the life. I just had to request him to build a video around it. Right after 5 days of my request, John created a wonderful video around this subject. I watched it multiple times as it was a wonderful video. I am not going to write about what was in the video much as I suggest you to watch the video itself. Here is one of the personal stories I want to share which is absolutely relevant to this video. I think my story 100% resonant the story of John. A Real Story from My Past Three years ago, I submitted a session in one of the SharePoint conference as a SQL Server session. My session was accepted and I prepared it very well. I put more than 2 month’s time to prepare for the session and I was very excited to present the session. I reached to the event place traveling thousands of the miles and I was very much excited to present the session. However, there was a little mixed up in the session. There were multiple session which were similar to my session title. One of the other speakers also had proposed a database related session and was selected. When the material went to print the printing team got confused and by mistake swapped the sessions. The other speaker got Performance with SQL Server session and I had received Performance with SharePoint session. IT was indeed a big mixed up but now that is how it was in the event guide and it was marketed the same way everything in the event. A Big Mix Up I had to talk with the event organizer and we come to the conclusion that we all had good intention but things just got mixed up and now was the time when “The show must go on“. I had a great amount of hesitation to go and present the session as I had personally never worked with Sharepoint so close in my life and my session abstracted talked about SharePoint tricks in depth. Two hours before the session I took the help of one of my friend and installed the SharePoint on my box. He showed me a few things here and there but it was never a good enough time to learn everything which I wanted to learn. The Moments of Confidence I was very scared and nervous to go on the stage as a SharePoint was not something I felt comfortable. However, I decided to go on stage with confidence as a SharePoint expert. Though I did not know SharePoint at the best, I had confidence that whatever I know is correct and I will not misguide people. I had no intention to fool people but I had no intention to accept that I am a fool and you all wasted your time and money to dedicate your time to attend my session. I decided to be honest but at the same time decided to take the session beyond my expertise. The sixty minutes of the session went very fine and I was able to manage all the difficult question at a satisfactory level. When the session was over my feeling was that I would have not presented or talked any different if I had more knowledge of the SharePoint at that time. I think it was one of my best sessions and it was reflected in the session feedback as well. I was the best speaker across all the track and my session had highest ranking. I was delighted and I learned a very valuable lesson. I must go beyond my limits and knowledge. I must aim higher and work harder. I should not lie but I should have confidence that I have a good heart and I put 100% in my efforts.  Lessions Learned Since this incident I have learned a lot about SharePoint and I am now a regular speaker at various SharePoint conferences along with SQL Server sessions. I am motivated and I am not afraid. I know people have lots of expectation from me but I have learned not to judge myself before I do my best. I leave the judgement of my efforts to my audience. I do not take the burden of the feedback on me, even though I know my audience have expected from me. I know what I know and I put my best. I must go out, if I fail, I learn from my mistake but I must keep my progress trajectory very high. As John said in the video, sometime success is not something we can achieve 100% but we can keep on going near to it. As long as we do not lose our focus from our goal and do not deviate from our progress path, we are doing things right. Reference: Pinal Dave (http://blog.sqlauthority.com)  Filed under: About Me, PostADay, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, T SQL, Technology

    Read the article

  • SQL SERVER – Find Most Active Database in SQL Server – DMV dm_io_virtual_file_stats

    - by pinaldave
    Few days ago, I wrote about SQL SERVER – Find Current Location of Data and Log File of All the Database. There was very interesting conversation in comments by blog readers. Blog reader and SQL Expert Sreedhar has very interesting DMV presented which lists the most active database in SQL Server. For quick reference he has included the size of the disk in KB, MB and GB as well. SELECT DB_NAME(mf.database_id) AS databaseName, name AS File_LogicalName, CASE WHEN type_desc = 'LOG' THEN 'Log File' WHEN type_desc = 'ROWS' THEN 'Data File' ELSE type_desc END AS File_type_desc ,mf.physical_name ,num_of_reads ,num_of_bytes_read ,io_stall_read_ms ,num_of_writes ,num_of_bytes_written ,io_stall_write_ms ,io_stall ,size_on_disk_bytes ,size_on_disk_bytes/ 1024 AS size_on_disk_KB ,size_on_disk_bytes/ 1024 / 1024 AS size_on_disk_MB ,size_on_disk_bytes/ 1024 / 1024 / 1024 AS size_on_disk_GB FROM sys.dm_io_virtual_file_stats(NULL, NULL) AS divfs JOIN sys.master_files AS mf ON mf.database_id = divfs.database_id AND mf.FILE_ID = divfs.FILE_ID ORDER BY num_of_Reads DESC If you like to read and practice with DMVs, I suggest to read the blog of my very good friend Glenn Berry. He is one DMV expert. Reference: Pinal Dave (http://blog.SQLAuthority.com) Filed under: SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, T SQL, Technology

    Read the article

  • SQLAuthority Book Review – DBA Survivor: Become a Rock Star DBA

    - by pinaldave
    DBA Survivor: Become a Rock Star DBA – Thomas LaRock Link to Amazon Link to Flipkart First of all, I thank all my readers when I wrote that I could not get this book in any local book stores, because they offered me to send a copy of this good book. A very special mention goes to Sripada and Jayesh for they gave so much effort in finding my home address and sending me the hard copy. Before, I did not have the copy of the book, but now I have two of it already! It surprises me how my readers were able to find my home address, which I have not publicly shared. Quick Review: This is indeed a one easy-to-read and fun book. We all work day and night with technology yet we should not forget to show our love and care for our family at home. For our souls that starve for peace and guidance, this one book is the “it” book for all the technology enthusiasts. Though this book was specifically written for DBAs, the reach is not limited to DBAs only because the lessons incorporated in it actually applies to all. This is one of the most motivating technical books I have read. Detailed Review: Let us go over a few questions first: Who wants to be as famous as rockstars in the field of Database Administration? How can one learn what it takes to become a top notch software developer? If you are a beginner in your field, how will you go to next level? Your boss may be very kind or like Dilbert’s Boss, what will you do? How do you keep growing when Eco-system around you does not support you? You are almost at top but there is someone else at the TOP, what do you do and how do you avoid office politics? As a database developer what should be your basic responsibility? and many more… I was able to completely read book in one sitting and I loved it. Before I continue with my opinion, I want to echo the opinion of Kevin Kline who has written the Forward of the book. He has truly suggested that “You hold in your hands a collection of insights and wisdom on the topic of database administration gained through many years of hard-won experience, long nights of study, and direct mentorship under some of the industry’s most talented database professionals and information technology (IT) experts.” Today, IT field is getting bigger and better, while talking about terabytes of the database becomes “more” normal every single day. The gods and demigods of database professionals are taking care of these large scale databases and are carefully maintaining them. In this world, there are only a few beginnings on the first step. There are many experts in different technology fields who are asked to address the issues with databases. There is YOU and ME, who is just new to this work. So we ask ourselves WHERE to begin and HOW to begin. We adore and follow the religion of our rockstars, but oftentimes we really have no idea about their background and their struggles. Every rockstar has his success story which needs to be digested before learning his tricks and tips. This book starts with the same note and teaches the two most important lessons for anybody who wants to be a DBA Rockstar –  to focus on their single goal of learning and to excel the technology. The story starts with three simple guidelines – Get Prepared, Get Trained, Get Certified. Once a person learns the skills, and then, it would be about time that he needs to enrich or to improve those skills you have learned. I am sure that the right opportunity will come finding themselves and they will not have to go run behind it. However, the real challenge for any person is the first day or first week. A new employee, no matter how much experienced he is, sometimes has no clue about what should one do at new job. Chapter 2 and chapter 3 precisely talk about what one should do as soon as the new job begins. It is also written with keeping the fact in focus that each job can be very much different but there are few infrastructure setups and programming concepts are the same. Learning basics of database was really interesting. I like to focus on the roots of any technology. It is important to understand the structure of the database before suggesting what indexes needs to be created, the same way this book covers the most essential knowledge one must learn by most database developers. I think the title of the fourth chapter is my favorite sentence in this book. I can see that I will be saying this again and again in the future – “A Development Server Is a Production Server to a Developer“. I have worked in the software industry for almost 8 years now and I have seen so many developers sitting on their chairs and waiting for instructions from their lead about how to improve the code or what to do the next. When I talk to them, I suggest that the experiment with their server and try various techniques. I think they all should understand that for them, a development server is their production server and needs to pay proper attention to the code from the beginning. There should be NO any inappropriate code from the beginning. One has to fully focus and give their best, if they are not sure they should ask but should do something and stay active. Chapter 5 and 6 talks about two essential skills for any developer and database administration – what are the ethics of developers when they are working with production server and how to support software which is running on the production server. I have met many people who know the theory by heart but when put in front of keyboard they do not know where to start. The first thing they do opening the browser and searching online, instead of opening SQL Server Management Studio. This can very well happen to anybody who is experienced as well. Chapter 5 and 6 addresses that situation as well includes the handy scripts which can solve almost all the basic trouble shooting issues. “Where’s the Buffet?” By far, this is the best chapter in this book. If you have ever met me, you would know that I love food. I think after reading this chapter, I felt Thomas has written this just keeping me in mind. I think there will be many other people who feel the same way, too. Even my wife who read this chapter thought this was specifically written for me. I will not talk any more about this chapter as this is one must read chapter. And of course this is about real ‘FOOD‘. I am an SQL Server Trainer and Consultant and I totally agree with the point made in the chapter 8 of this book. Yes, it says here that what is necessary to train employees and people. Millions of dollars worth the labor is continuously done in the world which has faults and incorrect. Once something goes wrong, very expensive consultant comes in and fixes the problem. This whole cycle which can be stopped and improved if proper training is done. There is plenty of free trainings available as well, if one cannot afford paid training. “Connect. Learn. Share” – I think this is a great summary and bird’s eye view of this book. Networking is the key. Everything which is discussed in this book can be taken to next level if one properly uses this tips and continuously grow with it. Connecting with others, helping learn each other and building the good knowledge sharing environment should be the goal of everyone. Before I end the review I want to share a real experience. I have personally met one DBA who has worked in a single department in a company for so long that when he was put in a different department in his company due to closing that department, he could not adjust and quit the job despite the same people and company around him. Adjusting in the new environment gets much tougher as one person gets more and more experienced. This book precisely addresses the same issue along with their solutions. I just cannot stop comparing the book with my personal journey. I found so many things which are coincidently in the book is written as how we developer and DBA think. I must express special thanks to Thomas for taking time in his personal life and write this book for us. This book is indeed a book for everybody who wants to grow healthy in the tough and competitive environment. Reference: Pinal Dave (http://blog.SQLAuthority.com) Filed under: Pinal Dave, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, SQLAuthority Book Review, SQLAuthority News, SQLServer, T SQL, Technology

    Read the article

  • SQL SERVER – Guest Post – Jonathan Kehayias – Wait Type – Day 16 of 28

    - by pinaldave
    Jonathan Kehayias (Blog | Twitter) is a MCITP Database Administrator and Developer, who got started in SQL Server in 2004 as a database developer and report writer in the natural gas industry. After spending two and a half years working in TSQL, in late 2006, he transitioned to the role of SQL Database Administrator. His primary passion is performance tuning, where he frequently rewrites queries for better performance and performs in depth analysis of index implementation and usage. Jonathan blogs regularly on SQLBlog, and was a coauthor of Professional SQL Server 2008 Internals and Troubleshooting. On a personal note, I think Jonathan is extremely positive person. In every conversation with him I have found that he is always eager to help and encourage. Every time he finds something needs to be approved, he has contacted me without hesitation and guided me to improve, change and learn. During all the time, he has not lost his focus to help larger community. I am honored that he has accepted to provide his views on complex subject of Wait Types and Queues. Currently I am reading his series on Extended Events. Here is the guest blog post by Jonathan: SQL Server troubleshooting is all about correlating related pieces of information together to indentify where exactly the root cause of a problem lies. In my daily work as a DBA, I generally get phone calls like, “So and so application is slow, what’s wrong with the SQL Server.” One of the funny things about the letters DBA is that they go so well with Default Blame Acceptor, and I really wish that I knew exactly who the first person was that pointed that out to me, because it really fits at times. A lot of times when I get this call, the problem isn’t related to SQL Server at all, but every now and then in my initial quick checks, something pops up that makes me start looking at things further. The SQL Server is slow, we see a number of tasks waiting on ASYNC_IO_COMPLETION, IO_COMPLETION, or PAGEIOLATCH_* waits in sys.dm_exec_requests and sys.dm_exec_waiting_tasks. These are also some of the highest wait types in sys.dm_os_wait_stats for the server, so it would appear that we have a disk I/O bottleneck on the machine. A quick check of sys.dm_io_virtual_file_stats() and tempdb shows a high write stall rate, while our user databases show high read stall rates on the data files. A quick check of some performance counters and Page Life Expectancy on the server is bouncing up and down in the 50-150 range, the Free Page counter consistently hits zero, and the Free List Stalls/sec counter keeps jumping over 10, but Buffer Cache Hit Ratio is 98-99%. Where exactly is the problem? In this case, which happens to be based on a real scenario I faced a few years back, the problem may not be a disk bottleneck at all; it may very well be a memory pressure issue on the server. A quick check of the system spec’s and it is a dual duo core server with 8GB RAM running SQL Server 2005 SP1 x64 on Windows Server 2003 R2 x64. Max Server memory is configured at 6GB and we think that this should be enough to handle the workload; or is it? This is a unique scenario because there are a couple of things happening inside of this system, and they all relate to what the root cause of the performance problem is on the system. If we were to query sys.dm_exec_query_stats for the TOP 10 queries, by max_physical_reads, max_logical_reads, and max_worker_time, we may be able to find some queries that were using excessive I/O and possibly CPU against the system in their worst single execution. We can also CROSS APPLY to sys.dm_exec_sql_text() and see the statement text, and also CROSS APPLY sys.dm_exec_query_plan() to get the execution plan stored in cache. Ok, quick check, the plans are pretty big, I see some large index seeks, that estimate 2.8GB of data movement between operators, but everything looks like it is optimized the best it can be. Nothing really stands out in the code, and the indexing looks correct, and I should have enough memory to handle this in cache, so it must be a disk I/O problem right? Not exactly! If we were to look at how much memory the plan cache is taking by querying sys.dm_os_memory_clerks for the CACHESTORE_SQLCP and CACHESTORE_OBJCP clerks we might be surprised at what we find. In SQL Server 2005 RTM and SP1, the plan cache was allowed to take up to 75% of the memory under 8GB. I’ll give you a second to go back and read that again. Yes, you read it correctly, it says 75% of the memory under 8GB, but you don’t have to take my word for it, you can validate this by reading Changes in Caching Behavior between SQL Server 2000, SQL Server 2005 RTM and SQL Server 2005 SP2. In this scenario the application uses an entirely adhoc workload against SQL Server and this leads to plan cache bloat, and up to 4.5GB of our 6GB of memory for SQL can be consumed by the plan cache in SQL Server 2005 SP1. This in turn reduces the size of the buffer cache to just 1.5GB, causing our 2.8GB of data movement in this expensive plan to cause complete flushing of the buffer cache, not just once initially, but then another time during the queries execution, resulting in excessive physical I/O from disk. Keep in mind that this is not the only query executing at the time this occurs. Remember the output of sys.dm_io_virtual_file_stats() showed high read stalls on the data files for our user databases versus higher write stalls for tempdb? The memory pressure is also forcing heavier use of tempdb to handle sorting and hashing in the environment as well. The real clue here is the Memory counters for the instance; Page Life Expectancy, Free List Pages, and Free List Stalls/sec. The fact that Page Life Expectancy is fluctuating between 50 and 150 constantly is a sign that the buffer cache is experiencing constant churn of data, once every minute to two and a half minutes. If you add to the Page Life Expectancy counter, the consistent bottoming out of Free List Pages along with Free List Stalls/sec consistently spiking over 10, and you have the perfect memory pressure scenario. All of sudden it may not be that our disk subsystem is the problem, but is instead an innocent bystander and victim. Side Note: The Page Life Expectancy counter dropping briefly and then returning to normal operating values intermittently is not necessarily a sign that the server is under memory pressure. The Books Online and a number of other references will tell you that this counter should remain on average above 300 which is the time in seconds a page will remain in cache before being flushed or aged out. This number, which equates to just five minutes, is incredibly low for modern systems and most published documents pre-date the predominance of 64 bit computing and easy availability to larger amounts of memory in SQL Servers. As food for thought, consider that my personal laptop has more memory in it than most SQL Servers did at the time those numbers were posted. I would argue that today, a system churning the buffer cache every five minutes is in need of some serious tuning or a hardware upgrade. Back to our problem and its investigation: There are two things really wrong with this server; first the plan cache is excessively consuming memory and bloated in size and we need to look at that and second we need to evaluate upgrading the memory to accommodate the workload being performed. In the case of the server I was working on there were a lot of single use plans found in sys.dm_exec_cached_plans (where usecounts=1). Single use plans waste space in the plan cache, especially when they are adhoc plans for statements that had concatenated filter criteria that is not likely to reoccur with any frequency.  SQL Server 2005 doesn’t natively have a way to evict a single plan from cache like SQL Server 2008 does, but MVP Kalen Delaney, showed a hack to evict a single plan by creating a plan guide for the statement and then dropping that plan guide in her blog post Geek City: Clearing a Single Plan from Cache. We could put that hack in place in a job to automate cleaning out all the single use plans periodically, minimizing the size of the plan cache, but a better solution would be to fix the application so that it uses proper parameterized calls to the database. You didn’t write the app, and you can’t change its design? Ok, well you could try to force parameterization to occur by creating and keeping plan guides in place, or we can try forcing parameterization at the database level by using ALTER DATABASE <dbname> SET PARAMETERIZATION FORCED and that might help. If neither of these help, we could periodically dump the plan cache for that database, as discussed as being a problem in Kalen’s blog post referenced above; not an ideal scenario. The other option is to increase the memory on the server to 16GB or 32GB, if the hardware allows it, which will increase the size of the plan cache as well as the buffer cache. In SQL Server 2005 SP1, on a system with 16GB of memory, if we set max server memory to 14GB the plan cache could use at most 9GB  [(8GB*.75)+(6GB*.5)=(6+3)=9GB], leaving 5GB for the buffer cache.  If we went to 32GB of memory and set max server memory to 28GB, the plan cache could use at most 16GB [(8*.75)+(20*.5)=(6+10)=16GB], leaving 12GB for the buffer cache. Thankfully we have SQL Server 2005 Service Pack 2, 3, and 4 these days which include the changes in plan cache sizing discussed in the Changes to Caching Behavior between SQL Server 2000, SQL Server 2005 RTM and SQL Server 2005 SP2 blog post. In real life, when I was troubleshooting this problem, I spent a week trying to chase down the cause of the disk I/O bottleneck with our Server Admin and SAN Admin, and there wasn’t much that could be done immediately there, so I finally asked if we could increase the memory on the server to 16GB, which did fix the problem. It wasn’t until I had this same problem occur on another system that I actually figured out how to really troubleshoot this down to the root cause.  I couldn’t believe the size of the plan cache on the server with 16GB of memory when I actually learned about this and went back to look at it. SQL Server is constantly telling a story to anyone that will listen. As the DBA, you have to sit back and listen to all that it’s telling you and then evaluate the big picture and how all the data you can gather from SQL about performance relate to each other. One of the greatest tools out there is actually a free in the form of Diagnostic Scripts for SQL Server 2005 and 2008, created by MVP Glenn Alan Berry. Glenn’s scripts collect a majority of the information that SQL has to offer for rapid troubleshooting of problems, and he includes a lot of notes about what the outputs of each individual query might be telling you. When I read Pinal’s blog post SQL SERVER – ASYNC_IO_COMPLETION – Wait Type – Day 11 of 28, I noticed that he referenced Checking Memory Related Performance Counters in his post, but there was no real explanation about why checking memory counters is so important when looking at an I/O related wait type. I thought I’d chat with him briefly on Google Talk/Twitter DM and point this out, and offer a couple of other points I noted, so that he could add the information to his blog post if he found it useful.  Instead he asked that I write a guest blog for this. I am honored to be a guest blogger, and to be able to share this kind of information with the community. The information contained in this blog post is a glimpse at how I do troubleshooting almost every day of the week in my own environment. SQL Server provides us with a lot of information about how it is running, and where it may be having problems, it is up to us to play detective and find out how all that information comes together to tell us what’s really the problem. This blog post is written by Jonathan Kehayias (Blog | Twitter). Reference: Pinal Dave (http://blog.SQLAuthority.com) Filed under: MVP, Pinal Dave, PostADay, Readers Contribution, SQL, SQL Authority, SQL Query, SQL Server, SQL Tips and Tricks, SQL Wait Stats, SQL Wait Types, T SQL, Technology

    Read the article

< Previous Page | 152 153 154 155 156 157 158 159 160 161 162 163  | Next Page >