We have a highly transactional database. It was owned by a third party before but now both the database and the application is on our site and we are trying to improve this project. So, we have a big (902919 rows), heap table, which is getting bigger and bigger everyday and sometimes deadlocks occur. The table has only 4 columns, "token", "type", "value" and "cacheTime", unique index cannot be created. It has one index on "token"(char(36)) and "type"(varchar(50)) ("value" should also be included but it is nvarchar(max)).
The driver table , which keeps track of what datamarts ran and for what date range gets updated frequently during the etl run . There can be as many as 250 updates issued on this table in a single second.
Now this table is a heap , and there are no indexes on it .
During these updates , we encounter deadlocks causing the ETL job to fail .
IF (SELECT OBJECT_ID('t1')) IS NOT NULLDROP TABLE t1GOCREATE TABLE t1 (c1 INT, c2 INT)DECLARE @n INTSET @n = 1WHILE @n <= 454BEGININSERT INTO t1 VALUES (@n, @n)SET @n = @n + 1ENDSELECT name, indid, CASE indidWHEN 0 THEN 'Table'WHEN 1 THEN 'Clustered Index'ELSE 'Nonclustered Index'END AS Type,dpages, rowcntFROM sysindexesWHERE id = OBJECT_ID('T1')name indid Type dpages rowcnt---- ----- ---- ------ ------NULL 0 Table 2 454I have a table containing 454 rows of two columnsof type INT with each being 4 bytesc1 int = 4 bytes+c2 int = 4 bytes=8 bytes per rowIf I entered 454 rows : 454 * 8 = 3,632 byteseach SQL Page is 8KB = 8 * 1024 bytes= 8,192 bytesa data page header takes the first 96 bytesleaving 8096 bytes for data and row offsets.Each record uses a row offset at the end of the pageconsisting of 2 bytes. 454 * 2 = 908 bytes.8096 - 3632 - 908 = 3,556 bytes. Should this befree data bytes?For a heap table, does SQL add an internal uniqueidentifiercolumn also? or my question is when does SQL adda uniqueidentifier? I am reading Inside SQL 2000 andtrying to understand a few things.A uniqueidentifier of 4 bytes gets added when a clustered indexexists but it is NOT a UNIQUE clustered index. AND onlyif duplicate record is added those two records only geta uniqueidentifier value.But in my example it's a heap table with no indexes. Evenon a heap table with no indexes a ROWID or Uniqueidentifierget added? Based on the INSERT statement above allvalues are unique.So what am I missing to understand why 453 rowsmake one data page to be used whereas 454 rowsmake two data pages to be used?Thank you
I have bunch of heap tables and the fragmentation seems to be high, i am not sure whether i shall add index for them, as these tables are inserted and updated every day.
We have some Deadlock alerts set up in SQL Agent that email us when the performance counter for deadlocks goes above zero. I've used the following script to identify the event file which has deadlock information in there.
select CAST(target_data as xml) as TargetData from sys.dm_xe_session_targets st join sys.dm_xe_sessions s on s.address = st.event_session_address where name = 'system_health'
Now that is fine, and we're looking into that (number of deadlocks appear to be 0.5) but out of interest ran a SQL Profiler session to capture the details as well and nothing is showing, I've received a few alerts and the trace file has information in there - but profiler shows absolutely nothing (all deadlock events are captured)
We are using partitions and all the table are properly aligned as per the partition keys. When this particular sp, which is inserting data to a table from a different table based on the partitionkeys is called by Web UI where threading has been applied, dead lock appears.
Let me make it more clear.
ThreadOne: Insert into table A(partitionKey,BatchId,...) select * from table B where partitionkey = 1 ThreadTwo: Insert into table A(partitionKey,BatchId,...) select * from table B where partitionkey = 2
I can see sometimes it gives deadlock for this procedure, not sure about the reason, as far as I guess since the tables are partitioned and escalation is set to Auto the deadlock should not occur.
I have a dts which creates a table which is utilized on my localintranet. The DTS runs without error and the table iscreated/populated/transfered to the appropriate db. Then it appearsthat there is an action on this table which truncates it. I have beenunable to determine the culprit. Can I create a trigger that willcapture truncation? I have tried to create a trigger to capture thisinformation but none that I attempt seem to work on capturing atruncation or a drop table and re-create.Any help would be greatly appreciated.MT.
In order to troubleshoot some deadlocking that I am seeing on SQL Server, I am trying to capture the Deadlock XML by enabling the Events Extraction Settings option 'Save Deadlock XML events separately' and specifying a Deadlock XML results file.
Meanwhile, I am also tracing the Deadlock graph, Lock:Deadlock, and Lock:Deadlock Chain events. Yet the xdl file remains empty even though I am getting hits on the events themselves in the SQL Profiler trace.
Also, I have the following trace flag settings enabled.
TraceFlagStatusGlobalSession 1204110 1222110
Why the xdl file remains empty even though (I think) it should contain some XML for deadlocks that are actually happening?
We are running SAP 3.1H on MS SQL server 6.5. We are having serius problems with printing. Here are some error messages from the system log.
Deadlock occurred DB error 1205 at UPD access to table TSP02 MSGNO 1205, MSGSTATE 2, SEVERITY 13 Your server command (process id 92) was deadlocked with another process and has been chosen as deadlock victim. Re-run your commandsrvname SAPPROD : procname YR200000015BKH4037TSP02 : line 2 Spool: RT_UPDATE error 1 for table TSP02 DB error 0 at UPD access to table TSP02
Spool: RT_UPDATE error 1 for table TSP02 DB error 0 at DEL access to table TST01
Database error: TemSe->XRTAB(5)->1 for table TST01 key [005]SPOOL25171 ,1 Database error: TemSe->XRTAB(5)->1 for table TST01 key [005]SPOOL25171 ,1 Some error in TemSe management DB error 0 at DEL access to table TSP01
The system log is full of these errors, we have a lot of inconsistency's in the temse objects. Can you tell me wat the solution is or cult be for this problem??
I need to discover the actual order in which locks are acquired on a table during a query.
This with a goal of analyzing the lock order of queries against the same table to prevent deadlocks.
I'm using SQL Server 2008 R2.
From Management Studio I execute:
begin transaction <my query> exec sp_lock rollback transaction
In the output I see interesting information about which locks are acquired, but:
- are this locks ordered by the time they're acquired? That is, can I be sure that lock at row n is acquired before lock at row n+1? - if not, how can I get this information?
We have a table that simply stores all changes to a specific record in another table and can get very large. The relationship is such there are many records in the transaction table for each record in the parent table. How many depends on how many times the record has been updated and can contain multiple entries for each column.
The transaction table contains a clustered index over a column that is defined as a UNIQUEIDENTIFIER. There are other indexes as well over business fields (basically foreign key columns). This obviously has some performance implications and becomes fragmented very quickly during heavy loads. Then as expected, the performance issues cascade to queries, etc.
Anyway, we are looking at two options - (a) removing the clustered index altogether and treat it as a heap or (b) add another column defined as identity (1,1) and make that the clustered index.
My initial research tells me the heap is not the way to go as there still may be performance issues with it. Using the second option guarantees me that all new data is always added to the end and will minimize the fragmentation. Keep in mind we do have regular maintenance jobs to rebuild / reorganize indexes and lob data.
Can anyone shed their thoughts on these two options for this situation?
Hi guru,I've been new company for only a month and started analysing IndexFragmentation.After I ran DBCC DBREINDEX and capture data into permanent table, I 'veseen lots of tables with no indexes. These tables showed:Very low scan density,High extent fragmentationHigh Avg. Bytes Free per PageWhat are the best strategies to defragment tables with no indexes?I'm planning to make a rule that each table must have a clustered indexand this index must be created on the best column (highestselectivity).Please help.Thanks,Silaphet,
I have a big table (heap)... well, not so big, I have a small serverand I want to spread access to it across several new disks dedicatedonly to that table.I known its possible to do that creating a clustered index with "ONfilegroup" option but I want to maintain it as a heap, is there anyway to do this without dropping indexes/references - bulk unload -create table - bulk load - create indexes?.
"The system reports 99 percent memory load. There are 8584744960 bytes of physical memory with 5799936 bytes free. There are 8796092891136 bytes of virtual memory with 8794956038144 bytes free. The paging file has 7447801856 bytes with 5201920 bytes free."
The packages running have been running for the last year with no issues. Admintidley they were the only jobs running against the Instance. I have now introduced additional databases and packages.
The package do not over lap when they run so there is currently no contention for resource.
I have no idea where to start looking to identify the culprit.
I have observed the available pagefile being completely consumed and physical memory. The only way I can get it released si by bouncing the instance.
the only thing that has changed is pagefile. The system has 8GB RAM and was configured with a 4GB pagefile. I have recently created a second pagefile on an alternative disk of 12GB and reduced the 4GB pagefile to 100MB and left it on the root drive. When I view the pagefile size in properties, it says that there is only 100MB even though it creates a 12GB pagefile.
1. Created a INSTEAD OF TRIGGER for INSERT 2. In the trigger, I have written a sql which inserts data in the same table 3. When I do the INSERT on the table the data in the trigger gets inserted 4. I was sort of expecting a recursive loop
Is this the normal behavior?
------------------------ I think, therefore I am - Rene Descartes
I have a simple invoice, inventory, billing program that is doing something strange but I can't track it down.
I can't figure out what triggers it but occasionally when I add an invoice to a customer all the other invoices are added to the last customer updated.
My update statement on the invoice table is
UPDATE Invoices SET Date = @Date, InvoiceTotal = @InvoiceTotal, SubTotal = @SubTotal, Tax = @Tax, CustomerID = @CustomerID WHERE (CustomerID = @CustomerID)
The only triggers on the table are to update the qty in Inventory and another trigger that fires when a payment is entered.
I have encountered a very frustrating situation when trying to use SQLBulkCopy. I have two excel files that I am trying to import into two tables in an MSSQL Server 2005 Express DB. One excel file has 5,000 rows, while the other file has 500,000 rows.I was able to import the smaller file successfully using this vb.net code: Protected Sub L26ExcelToSQL() 'Declare Variables Dim sSQLTable As String = "Local26Members" Dim sExcelFileName As String = "Full Local 26 List Formatted.xls" Dim sWorkbook As String = "[Sheet1$]"
Dim sSqlConnectionString As String = ConfigurationManager.ConnectionStrings("SiteSqlServer").ConnectionString.ToString 'Execute a query to erase any previous data from our destination table Dim sClearSQL = "DELETE FROM " & sSQLTable Dim SqlConn As SqlConnection = New SqlConnection(sSqlConnectionString) Dim SqlCmd As SqlCommand = New SqlCommand(sClearSQL, SqlConn) SqlConn.Open() SqlCmd.ExecuteNonQuery() SqlConn.Close() 'Series of commands to bulk copy data from the excel file into our SQL table Dim OleDbConn As OleDbConnection = New OleDbConnection(sExcelConnectionString) Dim OleDbCmd As OleDbCommand = New OleDbCommand(("SELECT * FROM " & sWorkbook), OleDbConn) OleDbConn.Open() Dim dr As OleDbDataReader = OleDbCmd.ExecuteReader() Dim bulkCopy As SqlBulkCopy = New SqlBulkCopy(sSqlConnectionString) bulkCopy.DestinationTableName = sSQLTable bulkCopy.WriteToServer(dr) OleDbConn.Close() End Sub However, when I tried to import the 500,000 row excel file, I got the following error: Server Error in '/L26' Application.
A transport-level error has occurred when receiving results from the server. (provider: TCP Provider, error: 0 - The specified network name is no longer available.)
Description: An unhandled exception occurred during the execution of the current web request. Please review the stack trace for more information about the error and where it originated in the code.
Exception Details: System.Data.SqlClient.SqlException: A transport-level error has occurred when receiving results from the server. (provider: TCP Provider, error: 0 - The specified network name is no longer available.)
[SqlException (0x80131904): A transport-level error has occurred when receiving results from the server. (provider: TCP Provider, error: 0 - The specified network name is no longer available.)] System.Data.SqlClient.SqlConnection.OnError(SqlException exception, Boolean breakConnection) +925466 System.Data.SqlClient.SqlInternalConnection.OnError(SqlException exception, Boolean breakConnection) +800118 System.Data.SqlClient.TdsParser.ThrowExceptionAndWarning(TdsParserStateObject stateObj) +186 System.Data.SqlClient.TdsParserStateObject.ReadSniError(TdsParserStateObject stateObj, UInt32 error) +556 System.Data.SqlClient.TdsParserStateObject.ReadSni(DbAsyncResult asyncResult, TdsParserStateObject stateObj) +164 System.Data.SqlClient.TdsParserStateObject.ReadPacket(Int32 bytesExpected) +34 System.Data.SqlClient.TdsParserStateObject.ReadBuffer() +44 System.Data.SqlClient.TdsParserStateObject.ReadByte() +17 System.Data.SqlClient.TdsParser.Run(RunBehavior runBehavior, SqlCommand cmdHandler, SqlDataReader dataStream, BulkCopySimpleResultSet bulkCopyHandler, TdsParserStateObject stateObj) +79 System.Data.SqlClient.SqlBulkCopy.WriteToServerInternal() +1336 System.Data.SqlClient.SqlBulkCopy.WriteRowSourceToServer(Int32 columnCount) +916 System.Data.SqlClient.SqlBulkCopy.WriteToServer(IDataReader reader) +151 _Default.CSVToSQL() in d:hostingmemberwolsite1L26DuesDefault2.aspx.vb:440 _Default.ButtonTest3_Click(Object sender, EventArgs e) in d:hostingmemberwolsite1L26DuesDefault2.aspx.vb:905 System.Web.UI.WebControls.Button.OnClick(EventArgs e) +105 System.Web.UI.WebControls.Button.RaisePostBackEvent(String eventArgument) +107 System.Web.UI.WebControls.Button.System.Web.UI.IPostBackEventHandler.RaisePostBackEvent(String eventArgument) +7 System.Web.UI.Page.RaisePostBackEvent(IPostBackEventHandler sourceControl, String eventArgument) +11 System.Web.UI.Page.RaisePostBackEvent(NameValueCollection postData) +33 System.Web.UI.Page.ProcessRequestMain(Boolean includeStagesBeforeAsyncPoint, Boolean includeStagesAfterAsyncPoint) +1746
Version Information: Microsoft .NET Framework Version:2.0.50727.1433; ASP.NET Version:2.0.50727.1433 After I received this error message, I tried viewing my database through the MSSQL Control Panel utilized by my hosting provider (WebHost4Life). However, I was unable to connect to the database and received this error: ___________________Microsoft OLE DB Provider for SQL Server error '80040e14' Database 1496 cannot be autostarted during server shutdown or startup. /getDBinfo.asp, line 29
_____________________ Now here is the most frustrating/mysterious part. I figured that maybe the error message were a result of the large size of the second excel file, so just for testing purposes, I created a new table in my MSSQL database. The table just has two fields, both set to varchar(50). I then created a test excel file, that had one row with the word "test" in the first and second column. When I tried using the code above to import the test excel data into the test table, I got the same exact error as I did with the 500,000 row file!Please help, I'm really stumped and I am not sure when I am having so much trouble replicating the success I had the 5,000 row file. Any suggestions are much apprecaited. -Bryan
Is there a way to either set Sql Server 2000 or ASP.net datetime fields to a standard format. The problem is that I am passing correct datetime fields using stored procedures and keep getting "Cannot convert datetime into string". It seems to me that many other developers are having that same problem. I tried alot of different methods and still have the same problem. I'm using c# and I never had a problem with datetime fields when I was using vb.net. The problem is that SQl Server is returning datetime formats that are not compatible with c#. I have code that works in other projects but when I try to use that same code I get that conversion error. How do I set the datetime in SQl Server and ASP.Net when I run queries so that the datetime output is in mm/dd/yyyy?
I get this mysterious error while running an update. SQLServer tells me that my update row is to big to fit in a row. Then I check the size of the row and and it is like 2000 bytes so it should fit without any problem at all. The really strange about it that when comments a field it works just fine. That field is always null. Could that be the problem. Anyone who have ever had a similar problem or so?
Second post for the day. I've just noticed a few indexes in our tables that have are named like such: hind_c_207_1. They seem to have been put in automatically, and are messing with our index structure. Anyone know where these come from and how I can stop them? Any help would be great.
Hey all,strange problem here... query #1 displays 357 records correctly and allis well. However, when placed within query #2 as a subquery, it updatesevery single record in the lta table, what's going on here? anythoughts?1.) select *from LTA INNER JOIN new_listON lta.voy = new_list.voy ANDlta.poe = new_list.poe2.)update ltaset lta.LL_RCVD = 'N'where exists (select *from LTA INNER JOIN new_listON lta.voy = new_list.voy ANDlta.poe = new_list.poe)
I am facing a pretty strange issue running a query in SQL Server through an ODBC client. Here is what is happening -
1. I am able to connect to SQL server. The connect entry is present in ODBC log as well as SQL Server trace. Running sp_who2 active indicates the connection to be active. No problem here.
2. At this point, running a simple query (something that returns in less than 10 mts or so) comes back with the result set. No problem here.
3. However, when a complex query (that it does not return anything even after 10 mts) is run, a run of sp_who2 active or dbcc inputbuffer commands indicate that the connection created initially is not there any longer. SQL Server trace indicates a logout event for the connection. However, ODBC log indicates that the connection is still active in SQL Server and the ODBC client continues to wait for the results of the query for hours until I kill it.
I use ODBC 4.2 client on AIX 5.2 and connect to SQL Server 2005.
One of the theories I have is that the AIX server does not respond to more than a certain number of keep alive messages resulting in SQL Server closing the connection gracefully thinking that the client is no longer available. However, even after changing the keep alive setting for TCP/IP in Network Configuration Menu to 4 hrs, the same behaviour continues.
Can someone throw some light as to what might be happening?
Hi All,I'm trying to track down a mysterious problem we're experiencing inwhich updates and inserts to tables in our mssql2k server appear to be'disappearing.'To explain our situation:We have a web page (written in ASP, if that's relevant) on which weaccept enrollment information.When that page is submitted, the form data is passed to a storedprocedure on our mssql2k server, which performs several operations,all of which are wrapped in a transaction.In particular, the stored procedure performs an update operation on arecord in one table (i'll call it TableA) and an insert into anothertable (TableB).If the procedure encounters a problem (ie after each update / insertoperation in the procedure we test for IF @@Error<>0) it performs arollback, performs a select similar to the one immediately below, andthen RETURNs.SELECT '1' as error, 'Unable to update TableA' as errormsgIf the procedure doesn't fail any of the @@Error tests, thetransaction is committed, and a membership number is SELECTed to bereturned.SELECT '0' as error, @memnum as membershipnumberThe @memnum variable is populated within the transaction.Back in the ASP page we test both for the proc returning an emptyrecordset, or for it passing an explicit value in the error field, andpush the page to an error page if either of these conditions are met.If, on the other hand, none of these conditions are met, and themembershipnumber field in the recordset is populated with a validmembership number, we push to a confirmation page.This confirmation page receives the membership number in a sessionvariable, performs a SELECT against TableB (the table that receivedthe insert during the proc) using that membership number in the WHEREclause, and the resultant recordset is used to populate theconfirmation details on that page. That recordset is also then used topopulate the details of a confirmation email, which is automaticallysent by the confirmation page.And now here's our problem: we've become aware of a handfull of peoplewho have gone through the enrollment process, have received theconfirmation email containing the information they supplied asexpected, but the data appears to be entirely missing from our tables.By that I mean that the record in TableA does not appear to have beenupdated (under normal circumstances that record should have hadseveral flags set, and several other fields updated with informationsupplied by the person enrolling), and the record in TableB does notappear to have been inserted.In essence, looking at our tables, it *feels* like the transaction inthe stored procedure for that particular enrollment hit a problem andwas rolled back. However, the evidence that we have in the form of theconfirmation email argues strongly that the data must have existed inour tables (particularly in TableB), if only for an unknown period oftime.We're kind of at our wit's end to work out what is going wrong withthese enrollments. From my understanding of transactions (and I couldwell be wrong) any changes to data (ie updates, inserts etc) containedwithin are essentially 'invisible' to any other operation (ie theSELECT that happens in the confirmation page) until the transaction iscommitted, implying that the effect of the update and insert shouldhave been 'permanently' successful if no error code is received and ifa valid membership number was returned. I ask, because someone in ourteam has suggested that maybe the operations in the transaction'lasted long enough' in the tables to have been visible for the SELECTon the confirmation page to have worked, but were then subsequentlyrolled back, explaining why the confirmation email is appropriatelypopulated and why the data then appears to be missing. However, as Isaid, this doesn't match my understanding of how transactions behave.Sorry for the length of this post, but I felt it was best to explainthis as best as I could.Does anyone have any advice they can give us on this situation? ie,are there any known problems with operations in transactions 'bleedingover' into tables, but then being rolled back at some later point?Does anyone have any thoughts or suggestions on how we can furtherdiagnose this issue?Truly, any help will be immensely appreciated...Thanks in advance,M Wells
Hi everyone,It looks like a mystery, but I hope there should be some explanation tothe issue I experience. Once in a blue moon a random stored procedurestops working the way it was designed. The stored procedure code looksunchanged. Recompiling, altering the code do not help. It looks like itis simply does not execute some part of it or does not run at all.However it returns no errors.One time a procedure entered into infinite loop and almost hang thewhole server.When I copy procedure code and save it under different name, it worksas designed. But nothing helps with existing procedure. The only wayhow to fix it is to completely drop and recreate.The problem is, that you usually have to do it in the middle of thebusiness day after you spent few hours trying to realize what wentwrong before you realize that you got another mysterious corruption. Ofcause I have no clue of how to detect such things in advance and toprevent them from occurring in the future.I can guarantee that the SQL code in those procedures was absolutelybug free, fully tested and was working fine for a long time.For the first time I thought that internal compiled code might corrupt.In this case altering or recompiling should help.I also thought about execution plan, but it should be also fixed bydoing things above.DBCC checkdb does not find any errorsThe issue never goes away until stored procedure is manually droppedand recreated with the same SQL code.So, I'm asking all if someone experienced something similar and canexplain how to prevent it, please share the knowledge. I wouldappreciate any type of help.Thank you.
Yesterday afternoon I lost access to my local MSDE 2000 sp3(a?)instance. Usinig NT or SQL authentication I get an 'access denied'error when I try to connect. This happened suddenly with no changesto the server on my part. I've tinkered with the Client NetworkUtility settings, from Multiprotocol only, every combination of Multiprotocol, TCP/IP, and named pipes. I've tried connecting with mynetwork machine name, using (local), and using my IP address (both127.0.0.1 and my network IP). I've tried connecting from QueryAnalyzer on other computers. Of course I've rebooted as well, all tono avail.I know that there were network changes being made, but my domain login(which I normally use to connec to my SQL Server as a sysadmin) isstill in the local administrators group on my machine and I haven'tnoticed any problems accessing anything else.Any ideas?
When executing some "big" update on a Sql Server 2005 standard edition SP1 (not clustered), sometimes it stops in a middle of work with this message:
The specified network name is no longer available. [SQLSTATE 08S01] (Error 64) Communication link failure [SQLSTATE 08S01] (Error 64)
I was reading some articles but they were all about clustered servers. Somewhere I read that it could be a memory problem, so we put 3G RAM and separate 20G paging partition beside the other data and backup partitions. The problem stays...
Funny thing is that it can sometimes happen but sometimes not with the same conditions!?
The server is Windows 2003 Server standard edition with SP1.
Hello, I have a report that looks fine in preview, but when put to PDF or printed contains gaps at the right and bottom.
Here is a picture of the problem: http://northeasttigers.webng.com/pdfproblem.jpg
My tables are the same width as the body, 15.5cm, and the report width is 21cm. Also adding the bottom table's top location attribute to it's height gives the same height as the body.
I get the following error messages in the sql server error log
Source Logon
Message Error: 18456, Severity: 14, State: 11.
and
Source Logon
Message Login failed for user 'NT AUTHORITYANONYMOUS LOGON'. [CLIENT: 185.23.11.33]
The scenario is: We set up log-shipping (LS) between a clustered sql server system (source server) and a stand-alone sql server box (target server). (SQL Server 2K5 EE + SP1), and LS goes very well, but on the target server, we found the above-mentioned error messages. BTW: the two servers are in the same domain.
