Monitor (Check Indexes) taking up CPU time

marcos1 · Post by **marcos1** » Tue Sep 30, 2003 6:23 pm

Hello,

We have a large webinator DB (over 1.4 million rows) and have recently started experiencing a problem where Monitor (Check Indexes) takes up 50% or more of the CPU time for extended periods of time (most of the time, in fact)-- slowing down everything else the server is trying to do. This only started in the last few days and we have not added any rows for several weeks.

Is this normal behavior for large tables with big indexes or might something be awry?

Thanks,

Evalumetrix

marcos1 · Post by **marcos1** » Tue Sep 30, 2003 8:31 pm

There are a number of files in the index directory that start with T15788a including one that end in _T.btr (62K) and a table of the same name (.dat). In monitor.log there is an error trying to find a table that it looks like was manually deleted by mistake (instead of dropping) -- I went ahead and dropped it. Can we just delete the T15788a stuff (it is not listed in SYSTABLE or SYSINDEXES)?

Thanks,

Evalumetrix

Post by **mark** » Wed Oct 01, 2003 10:49 am

Yes, if it's not listed in SYSINDEX you can delete it.

marcos1 · Post by **marcos1** » Wed Oct 01, 2003 11:01 am

OK. We deleted those file and we dropped and started recreating all indexes -- which seemed to fix things -- but when it got to rebuilding one of the indexes (xhtmlid) everything slowed to a stop. The machine actually kicked us out of the ssh session -- but right before that I could see that Monitor (the primary process) was taking up 99.9 percent of the CPU. Could this indicate that the table itself has some corruption? Other ideas?

Thanks,

Evaluemtrix