Ad Widget

Collapse

oracle db backend issues

Collapse
X
Collapse
 
  • Page of 1
  • Time
  • Show
Clear All
new posts
Previous template Next
  • username_taken
    Junior Member
    • Oct 2009
    • 11
    #1

    oracle db backend issues

    Using Oracle ( 11.2.0.1.0 ) as the DB backend for Zabbix ( 1.8.3 ) .

    We're seeing the occasional crash/hang where the oracle shared pool seems to get locked up on a query and stop processing queries. This fills the system up quickly and load average goes through the roof.

    We can fix it by restarting oracle, or by flushing the shared pool via 'alter system flush shared_pool;' in oracle.

    Obviously this isn't an ideal situation. Has anyone else seen this and been able to resolve it?
    Tags: None
    • untergeek
      Senior Member
      Zabbix Certified Specialist
      • Jun 2009
      • 512
      #2
      We are using 11.2.0.2. We had a strange blip right after we upgraded, but nothing since. I will keep my eyes out.

        Comment

        • jalexandre
          Junior Member
          • Jul 2010
          • 15
          #3
          Zabbix Strange Issues

          Talking about Oracle, we have some issues with Zabbix 1.8.3 and Oracle 10g:

          - Huge load averages when configuring Maps.
          - High Steal Time in Physical server (IBM P520 SMP 2 physical cores, 4 Virtual Cores).
          - Locked Process taking 100% of all CPUS, splited in User and Steal Times.


          Someone facing similar problem?

            Comment

            • untergeek
              Senior Member
              Zabbix Certified Specialist
              • Jun 2009
              • 512
              #4
              We're seeing the occasional crash/hang where the oracle shared pool seems to get locked up on a query and stop processing queries. This fills the system up quickly and load average goes through the roof.

              We can fix it by restarting oracle, or by flushing the shared pool via 'alter system flush shared_pool;' in oracle.

              Obviously this isn't an ideal situation. Has anyone else seen this and been able to resolve it?
              We've seen this problem twice now since we upgraded from 11.2.0.1 to 11.2.0.2. This is what our DB expert did:

              I checked Oracle support site for any related issues for this – there are few “work-around” approaches mentioned to address the “memoryleak”/”fragmentation” of shared pool area. I have put those in place – let us see if it helps.

              If not, there are few more “tuning” related parameters we can try next time we end up bouncing the database.

              Most likely the issues are coming from “not using bind variables” – which we know is not going to be addressed (at least not in the near future).

              As you know already, we had to put a job to flush the shared_pool from time to time (every 10 min. actually) and that may be “fragmenting” the shared pool according to Oracle – but no data I could see in the DB to be sure of it.
              I am not sure what these "work-around" approaches might be, but I can ask if you're interested.

                Comment

                Previous template Next
                Working...