i believe that all that witango start and stop transactions are is just sending "transaction start" and "transaction stop" to the database, not doing anything fancy, complicated or specific to witango.
----- Original Message ----- From: "Mike Scally" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Friday, December 10, 2004 2:21 AM Subject: RE: Witango-Talk: URGENT: MS-SQL hanging - process non-yielding Dave, Do you use transactions at all in your development? A while back I had some problems where queries were not getting executed fully, the Enterprise Manager wouldn't respond and the likes and it was due to Witango transactions (and possibly the way they were used). I then noticed the Witango had 'killed' in the log after a period of time on the file that was executing the query. Just a thought. Mike. -----Original Message----- From: Dave Machin [mailto:[EMAIL PROTECTED] Sent: 09 December 2004 20:26 To: [EMAIL PROTECTED] Subject: Re: Witango-Talk: URGENT: MS-SQL hanging - process non-yielding One new development -- we started to systematically disable .taf applications by putting 'temporarily unavailable' messages in them and may have found the one that is causing the problem. On our development machine it's always worked fine - but one thing unique about it is the size (in characters) of the query it executes. The query can get up to 150KB in size - the strange thing that happens is that in the debug mode WiTango doesn't display the query that it executes. The query's always worked in development though and returns in about 30 seconds. Is there some kind of limit to the size (how many characters) a query can be and remain stable? Dave ----- Original Message ----- From: "Robert Shubert" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Thursday, December 09, 2004 11:34 AM Subject: RE: Witango-Talk: URGENT: MS-SQL hanging - process non-yielding > That's very strange. What is your datasourcetimeout set to? I see that > message about once every 10-15 minutes, depending on a lot of variables. > Witango should have no need to open more than about 5 connections to any > one datasource on all but the busiest databases. It should also only be > able to open one per thread up to threadpoolsize. > > Make sure you're not getting flooded with requests from somewhere. Have > you restarted MS SQL? I have seen what you are describing, but at a rate > of about once a year, which I can write-off as a random incident. > > More times than not, you'll run into this kind of situation when you are > using Transactions, are you? > > Lastly, you might want to check over the indexing of the tables. I'm not > certain that this is possible on SQL 2K, but on an older version I once > made 2 clustered indexes on the same table (accidentally) and SQL really > didn't appreciate that much (but let it happen!). Poor indexing could > cause inserts/updates to take a tremendous amount of time. > > While you have the SQL server responding to you in EM, check the Current > Activity under management, and see if you connections (and their states) > seem appropriate. > > If you think you have a very long running db request which is causing > the problem, you can try using the governor to catch it. > > That's about all of my advice. Oh, you might want to set profiler going > and just watch what's happening, it should show you the statements right > before the stall. > > Robert > > -----Original Message----- > From: Dave Machin [mailto:[EMAIL PROTECTED] > Sent: Thursday, December 09, 2004 1:34 PM > To: [EMAIL PROTECTED] > Subject: Re: Witango-Talk: URGENT: MS-SQL hanging - process non-yielding > > None of the queries in the log look unusual. We reran many of them and > they > all return fine. We did see a query timeout in the witango events log > and > the strange thing was that if we subtracted the timeout seconds from the > timestamp of the event and looked back in the log we didn't find an > entry > for any query related to the application that timed out. > > We're also getting INFO messages at what I would think to be an alarming > rate saying 'no existing connection to the data source found, creating a > new > connection' - they're being written about 1 per second all the time. In > fact, just now, WiTango was writing those message so quickly that I > couldn't > open the events log file... > > ----- Original Message ----- > From: "Bill Conlon" <[EMAIL PROTECTED]> > To: <[EMAIL PROTECTED]> > Sent: Thursday, December 09, 2004 9:52 AM > Subject: Re: Witango-Talk: URGENT: MS-SQL hanging - process non-yielding > > > > Can you inspect the queries in the Witango.log to look for the > deadlock? > > > > On Thursday, December 9, 2004, at 09:30 AM, Dave Machin wrote: > > > > > Has anyone seen the following situation: > > > > > > Once or twice a day our MS-SQL SP3 database server stops responding. > > > All > > > jobs on it hang and it is unresponsive to enterprise manager. > WiTango > > > locks > > > up waiting on queries to return. The database log shows several > > > occurences > > > of the following error: > > > > > > Process 69:0 (6b0) UMS Context 0x0D37A8B0 appears to be non-yielding > on > > > Scheduler 2. > > > Error: 17883, Severity: 1, State: 0 > > > > > > We've installed a recommended hotfix from Microsoft to no avail. > > > > > > When we kill the WiTango service the database comes back. > > > > > > Since we can't connect to the database during this time, I can't > tell > > > what > > > the process ID mentioned in the error is - by the time we can get > back > > > in > > > it's gone. Since killing WiTango seems to fix it, I'm assuming it's > a > > > WiTango query of some kind. > > > > > > We're running: W2K SP4, WiTango .065 with SQL Server ODBC driver > > > version > > > 2000.85.1022.00. Database is W2K SP3, MS-SQL SP3. > > > > > > Any help would be appreciated... > > > > > > Dave Machin > > > > > > > > > > _______________________________________________________________________ > > > _ > > > TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf > > > > > > > > ________________________________________________________________________ > > TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf > > > > > > > ________________________________________________________________________ > TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf > > ________________________________________________________________________ > TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf > > > ________________________________________________________________________ TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf ******************************************************************** This message is intended only for the use of the person(s) ("the intended recipient(s)") to whom it is addressed. It may contain information which is privileged and confidential within the meaning of applicable law. If you are not the intended recipient, please contact the sender as soon as possible. The views expressed in this communication may not necessarily be the views held by LGCSB (Local Government Computer Services Board). Any attachments have been checked by a virus scanner and appear to be clean. Please ensure that you also scan all messages, as LGCSB does not accept any liability for contamination or damage to your systems. ******************************************************************** ________________________________________________________________________ TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf ________________________________________________________________________ TO UNSUBSCRIBE: Go to http://www.witango.com/developer/maillist.taf
