Hello Clovis, each database session will need its own 'User Task'. So check your application whether it forgets to close connections, and keep them hanging around. Each 'User Task' will also need shared resources, some of them only if the load increases, some of them from beginning.
See a previous mail at http://listserv.sap.com/pipermail/sapdb.general/2002-October/008942.html on what you may do (as it can be assumed that 150 MAXUSERTASKS will blow the setting, and make the kernel crash). BTW, usually not only the 'suspect' lines but the overall context in the knldiag is of interest - it may tell what happened around the time the error appeared, if there is one, also the possible stack backtrace in knldiag in the event of a crash (or the note of the absense of a stack backtrace ...). Anyway, if your application is loosing connections by forgetting them, no value of MAXUSERTASKS is sufficient to cover it over the time ... so try to use the appropriate client trace (JDBC/ODBC/Precompiler) to check where and how often the application does open and close connections. Regards Alexander Schr�der SAP DB, SAP Labs Berlin > -----Original Message----- > From: Cl�vis Wichoski [mailto:[EMAIL PROTECTED] > Sent: Tuesday, December 09, 2003 1:34 PM > To: [EMAIL PROTECTED] > Subject: Doubts about USER TASKS > > > Hi, > > I'm using SapDB 7.3.0 Build 040-000-091-740 on RedHat Linux 9. > > Here I'm having troubles with user tasks in production, in any time > user tasks goes quickly to top of MAXUSERTASKS and then the > server stops > to respond for new COMMUNIC threads, the only way is to > change database > to off-line then to warm again, this is a reinstalation after > a server > HD crash, but the RedHat Linux and SapDB are the same versions of old > crashed instance. > > I followed a message (http://lists.mysql.com/maxdb/19368) sended by > Markus D�hr, and setup my MAXUSERTASKS from 50 to 100 but the > user tasks > again goes quickly to top of MAXUSERTASKS. > ps: If I attempt to change this to 150 the server don't start and in > knldiag the message appear: > ERR 11330 COREHAND ABORTING due to signal 11 > > Now I need to solve this, and make sure that this problem don't occur > again. > I have a suspect case, checking the knldiag I have follow messages: > > MsgID: 11509 > Label: COMMUNIC > Message-Text: Connection broken by appl state 10 T74 > > MsgID: 53214 > Label: B*TREE > Message-Text: SET READ ONLY: 1A00 FFFFFFFFFFFF0605 > > What this messages explain? > What is the possible ways to USER TASKS increase quickly? > There is way to check what a USER TASK it is executing? > > another tips? > > Thanks for any help > > Clovis > > > -- > MaxDB Discussion Mailing List > For list archives: http://lists.mysql.com/maxdb > To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED] -- MaxDB Discussion Mailing List For list archives: http://lists.mysql.com/maxdb To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
