Solman crashes

[shradhasampat]

Hello,

Solman crashes every two week or so.

1. I noticed this error just before the crash in sm21 of solman sid1 for managed system sid2:

ORA-00020: maximum number of processes (500) exceeded

We have already increased processes at DB level to 1000,but  slowly  all processes gets occupied.

2. This error usually occurs in sm21 of solman SID1 for managed system SID2 when SID2 is shutdown on daily basis for backup.

Database error 1017

Database error 1017 at CON

> con=2(mgm22:1527/SID2): ORA-01017: invalid username/password;

> logon denied

From analysis we noticed that there is a job EFWK RESOURCE MANAGER (this is executed every min ) which tries to connect managed systems & this is the reason why we get above error on daily basis in sm21 of SID1.

1. Since this job attempts to connect to SID2 no. of times when SID2 is actually unavailable, the job then slowly occupies all db processes & these processes are never released. Could this be the reason for solman crash ?

2. I understand that we cannot restrict execution of job EFWK RESOURCE MANAGER to run only for particular duration unless we use external scheduler or change standard program but would like to know is it necessary to run this job & that too for every 1 min ?

3. Is there a workaround for this situation ?

Thanks

Shradha

[Former Member]

Hi Shradha,

     As you said , ORA-00020 occurs when all the db processes are occupied for any activity.

1. I noticed this error just before the crash in sm21 of solman sid1 for managed system sid2:

ORA-00020: maximum number of processes (500) exceeded

We have already increased processes at DB level to 1000,but  slowly  all processes gets occupied.

Increase to more number of processes then may  be to 2000... Check what are the DB activities are going on while backup is going on ... How many parallel backup threads are input.

http://nimishgarg.blogspot.in/2012/05/ora-00020-maximum-number-of-processes.html

2. This error usually occurs in sm21 of solman SID1 for managed system SID2 when SID2 is shutdown on daily basis for backup.

Database error 1017

Database error 1017 at CON

> con=2(mgm22:1527/SID2): ORA-01017: invalid username/password;

> logon denied

Login to SID2 system, execute R3trans -d and check what is the return code... If it  0000 , then there is a problem with the password updated in your SID1 for SID2... Kindly check that....

From analysis we noticed that there is a job EFWK RESOURCE MANAGER (this is executed every min ) which tries to connect managed systems & this is the reason why we get above error on daily basis in sm21 of SID1.

1. Since this job attempts to connect to SID2 no. of times when SID2 is actually unavailable, the job then slowly occupies all db processes & these processes are never released. Could this be the reason for solman crash ?

Please check in your SID1 system how many processes set... If possible, try to increase that also.. It doesnt mean that because of the SID2 being down could be the reason.

2. I understand that we cannot restrict execution of job EFWK RESOURCE MANAGER to run only for particular duration unless we use external scheduler or change standard program but would like to know is it necessary to run this job & that too for every 1 min ?

Please check the note Note 894279 - Background processing in SAP Solution Manager on details why this job being executed.. It is explained in a very good manner...

3. Is there a workaround for this situation ?

Please try to increase the processes in both SID1 and SID2 to 2000 .. Please see the above link on calculation for increasing the parameters.

I hope this information helps.

Thanks and Regards,

Vimal

[shyam_dontamsetty]

Hi Shrada,.

Since this job attempts to connect to SID2 no. of times when SID2 is actually unavailable, the job then slowly occupies all db processes & these processes are never released. Could this be the reason for solman crash

Have you checked any abnormalities in alert log? I dont think resource unavailable will cause and hold up the process in oracle. You need to look for the activites around the crash time in ora alert log. if possible could you give  us the log details around the crash time.

Best Regards,

Shyam

[former_member188883]

Hi Shradha,

How many Solution Manager instances you are talking about ?

What is SID1 and SID2 ?

As you mentioned you have increased the number of processes but have you modified the number of sessions as well  in Oracle ?

Regards,

Deepak Kori

[shradhasampat]

Thanks Deepak for replying !!

SID1 is solution manager system

SID2 is managed system

I think once we increase processes parameter the sessions parameter gets auto calculated.

[former_member188883]

Hi Shradha,

I think once we increase processes parameter the sessions parameter gets auto calculated.

IT IS NOT. You need to adjust the number of sessions in initSID.ora and create a new spfile.

If you have not done so, please do it and post the results after the changes.

Regards,

Deepak Kori