Locked SMQR queues when High load of idoc messages

[Former Member]

Dear all forum users,

We have the following simple scenario :

ECC 6.0 > Idoc MATMAS (reduced one) > XI > XML Third party tool format.

In ECC 6.0 creation of idoc depends on change pointer mechanism. In case of mass modification , a lot of messages can be created and in this kind of high load period we have a lot of errors that occur in processing RFC queue in our central Integration engine.

Error found are :

- "ThCPIC: ThICommit / connection closed (no data)"

- "COMMIT/ROLLBACK error on "DEFAULT" database connection"

It is not a fatal error since the scheduling of RSQIWKEX program restart queues with error automatically but is there a way to avoid these errors ?? Is it a pb of tuning ?

Thanks a lot.

Jean-Charles

[Former Member · Accepted solution]

hi...

try this blog:

regards,

Sreedhar Goud.

[Former Member · Accepted solution]

Thanks. I already know this blog.

[SudhirT]

Hi Jean,

I dont think its performance related problem specific to XI Bcoz in that case simply the number of trfc entries increases and consequently the queue entries smq1/smq2 increases without errors with status transaction recorded.

But as this is giving some error +"COMMIT/ROLLBACK error on "DEFAULT" database connection" +

If this error is frequently seen then it might be tablespace related problem in Database. Check the tablespace in DB02 for PSAPSR3 and PSAPSR3700 if these two have sufficient free space .

thanks!

[Former Member]

Thanks Sudhir,

I see with my admin and let you informed.

What are the corresponding data of these both tablespaces ?

[SudhirT]

Hi Jean,

Tablespaces contains altogether different data like PSAPSR3 will always contain transactional data and PSAPSR3700 will contain data related to applications like info related to your patch.

Thanks!

[Former Member]

Administrators have added Oracle DB sessions and DB process max number.

And it is now OK