SELECT FROM SVERS

[nicola_blasi]

hello

after a reorg of some tablespaces of a 46c instance with oracle 10.2.0.4 i have some strange things in st04.

I have many active processes like:

22227 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

21903 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

22026 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

22231 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

21899 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

21897 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

22227 ACTIVE PX Deq: Execution Msg SAPR3 SELECT FROM SVERS #

and so on....

the corresponding process in O.S is an ora_pxxx_dbsid

if I do a truss -p of this process i have :

truss -p 22227

semtimedop(33554613, 0xFFFFFFFF7FFF4F94, 1, 0xFFFFFFFF7FFF4F80) (sleeping...)

semtimedop(33554613, 0xFFFFFFFF7FFF4F94, 1, 0xFFFFFFFF7FFF4F80) Err#11 EAGAIN

semtimedop(33554613, 0xFFFFFFFF7FFF4F94, 1, 0xFFFFFFFF7FFF4F80) (sleeping...)

semtimedop(33554613, 0xFFFFFFFF7FFF4F94, 1, 0xFFFFFFFF7FFF4F80) Err#11 EAGAIN

times(0xFFFFFFFF7FFF5200) = 1037852813

times(0xFFFFFFFF7FFF5150) = 1037852813

semtimedop(33554613, 0xFFFFFFFF7FFF4F94, 1, 0xFFFFFFFF7FFF4F80) (sleeping...)

semtimedop(33554613, 0xFFFFFFFF7FFF4F94, 1, 0xFFFFFFFF7FFF4F80) Err#11 EAGAIN

Sap continues to work without problems anyway i don't understand this issue.

Thanks

Nicola

[stefan_koehler · Accepted solution]

Hello Nicola,

the wait event "PX Deq: Execution Msg" is an idle event per definition.

http://download.oracle.com/docs/cd/B19306_01/em.102/b25986/oracle_database.htm#sthref949

It seems like you have reorganized the tablespaces with the PARALLEL clause, but you did not change it back. So now you are using parallel query for these reorganized objects.

If this is a wanted behaviour it's ok ... if not please revert the parallelism of the corresponding objects. Refer to sapnote #651060 - subpoint 9 and 10 for changing the parallelism.

Regards

Stefan

[nicola_blasi · Accepted solution]

Hy Stefan

You are right...i've put more than 5000 indexes that don't change back to degree 1.

Thanks

Nicola