Oracle 12c on redhat SLOWWWWWW

[RAMON FERRERIS]

Hi:


I opened an issue with support, but would like to ask.

Have any one found a problem were running oracle 12c on redhat  the autorep commands are very very slow,  I have only 8 thousand jobs  and it takes up to 3 hours to do an autorep -j ALL -q on the system.     Scheduler and database are on 2 different red hat servers.  

Thanks for any ideas.

Database is under 400 megs

[Yehonatan Amir]

Hey Ramon,

We are using autosys with Oracle 12c on Redhat.. had no performance problems whatsoever

We also have a couple thousands  of jobs (not sure to exact number..)

Did you check if there is any irregularity with the server or DB load ? 
You could monitor the DB to see if the the query is taking very long , so you can tell if the delay is DB related or something on the app server

Good luck :)

[Steve Carrobis]

turn cursor sharing to FORCED. also there is a fix. I discovered this last year with Al Polanco. 12cR1 and SP6CUM1. there is a fix (T54Y090 ) and SP8 should also have the fix. 
Good Luck.

------------------------------
Steve C.
------------------------------

[GREG BRADLEY]

Hi Ramon,

I had an issue last year with Oracle 12.c on RHEL 6.9 with WAAE r11.3.6 sp6 CUM1 scheduler where after Oracle Admin Client upgrade from 11.2 to 12c the scheduler started taking core dumps and crashing. Refer to the following Resolution posted to my CA Support case. We have 120,000+ jobs defined and running up to 240,000 batch cycles per day so it is a fairly busy system. 

I don't know if this could be related to your slowness but the following is something for you to check.

RESOLUTION
RCA done by CA was a known issue with Oracle 12c and glibc < 2.14 (Linux Only) Oracle 12c Client Memory Leak Oracle 12c client libraries have a known memory leak when the glibc version on the machine is below 2.13. The root cause of the memory leak is due to a bug in GNU glibc that is related to a multi-thread program that uses dlopen() to load the libraries. To resolve this issue, apply glibc 2.14 or higher. Contact your Linux vendor to get the GNU glibc compatible patch for bug 12650 based on your Linux operating environment. But we cannot upgrade our gblic to 2.14 as we are running Red-hat 6 for which which glibc has rolled up at 2.12. The only option recommended by CA was to downgrade Oracle client to 11g and this was done and stablity returned to the system.

We have been on 11.2 Ora Client since then and will go back to Oracle 12c or higher once we get onto RHEL7.n with the glibc package at 2.14 or higher.

There are also known issues with autorep -J % slowness that have been re-mediated in r11.3.6 sp7 but not sure if lingering issue. 

From the RELEASE NOTES:

29.AUTOREP -J % COMMAND TAKES LONGER TIME TO
EXECUTE

The autorep -j ALL command fetches all the job
runs and displays the results faster than the
autorep -j % command.

Problem Number: 4742
_______________________________________________________________________________

Regards
Greg

[RAMON FERRERIS]

Thanks for the info, Iam waiting on Broadcom  support

 

[Steve Carrobis]

There's a fix for SP6CUm1 or you can use SP8.
These were all discovered by yours truly and all he necessary changes to autosys were made by Al polanco
Good luck.

Steve C.

[RAMON FERRERIS]

Thanks for the information, I will pass it to Oracle and respective groups.

 

Thanks

RF