Hello all,
We have many instances on an Oracle RAC cluster with two nodes. The check “Check_MK” very often fails because it gets no results or the execution time is too long (CRIT - [agent] keepalive timed out CRIT, Got no information from host, execution time 300.0 sec).
I was able to limit the behavior to the oracle_tablespaces plugin. The SQL statement executed there takes more than 10 minutes on one instance.
Some of the databases have many tablespaces (15+) with several data files.
Is there any way to shorten the check for the tablespaces?
The check_mk version we use is 1.6.0p13.
Hi Mathias,
mostly this is relatet to performance issues inside Oracle:
disable recyclebin
The SQLs on dba_free_space are extremly slow in some situations. Oracle released some bugfixes but the best solution ist disabling the reyclebin and purge all objects after then.
dictionary Statistics
Refresh the Statistics inside the Data-Dictionary of Oracle. They are not automatically refreshed in older versions of Oracle.
Execute: dbms_stats.gather_dictionary_stats
That solved most of all performance issues with the Tablespace-Check at our customers.
thank you for your answer.
I disabled the recyclebin, deleted all objects and recalculated the dictionary stats.
Unfortunately, the problem is still there.