since two days we´ve an running backup without any work. Zero bytes are backed up at the momtent. The system crashed after the backup was started scheduled.
For the moment we cant force the application we cant kill -9 the running process and the application is in state "request interupted".
If a db2 process cannot be killed by kill -9, then it means that the process is hung in the os level and OS support needs to find the root cause and fix the problem. Are there any entries in the operating system logs from the time of the crash just after the backup had started?
I´m not sure whether this is os relevant. When I kill -9 the processes db2bp they come back after five secounds or so. All my tablespaces are in state 0x0800 so how can I set the state to normal. With the switch command to one tablespace nothing was changed.
db2bp is a one kind of background process for the DB2 CLP (DB2 Command Line Processor). It is the process which actually connects to the database. DB2 CLP allows OS as well as DB2 commands/statements, this background process is required.
Please refer this useful link to get more information [about DB2 UDB Processes|http://www.ibm.com/developerworks/data/library/techarticle/0304chong/0304chong.html].
db2med and db2bm processes will play an important role while Backup operation in progress.
If your SAP System is started/running and you are able to access SAP System, you can check the running Background Job of Backup (starting from DBA*), If you have scheduled it using DB13. Cancel active Background Job from SM37.
If still that background job is running and if it is visible in SM50, you can cancel it by choosing "Process-> Cancel without Core" for the selected Background Job. If still its not canceled, You can cancel it from OS level, by getting valid Process ID details from SM50.
As final way, If your SAP Instance is not started/running, then you can execute db2stop and then db2start to restart the db2 instance on the host.
Bhavik G. Shroff
the job is started via script from a backup server not via db13. The sap is up and running and works fine. I only will delete the suggested running backup which nothing done and takes the tablespace in state 0x0800.
I get a backup information when I use the command "db2 list utilities show detail" wihich is started at 02.02.2010 6pm and backed up 0 bytes.
For the moment I cant restart the database because this is a high available system used 7x24. So it sould be another way to cancel the backup an restart an new backup.
Can you check if the backup process is waiting on a "latch" or whether there are any dead latches? You can check this with db2pd -latches. The output should be pretty transient and you will most likely not see the same latches when you run this command multiple times. Also you can check the output of that command to see whether any latches are held by the agent running the backup.
If this is a dead latch issue then the only way to cancel it will be to recycle the instance.
I have an OSS/PMR open for the exact same latch which causes our stmm process to hang similar to your backup process (our backups are done using flash so that doesnt hang). Update so far was that there is no way to interrupt or cancel the process without restarting and a permanent fix might be provided in a later FP.
Hope this helps!