rhq_availability and purging

Elias Ross genman at noderunner.net
Mon Jun 30 17:56:30 UTC 2014


I have over 20 million rows in rhq_availability.

One reason is I don't purge this data. The purge process for very
large tables (traits for example) usually removes a lot of rows, which
impacts inserts, which in turn impacts the agents through timeouts.

I'd like it if the trait data could move to Cassandra.

But I'd also like to have rhq_availability move as well to Cassandra.

If not, then how about changing the purge process to the following:
1) Select the rows needed for deletion
2) Then in a separate transaction, delete 100 at a time.


More information about the rhq-devel mailing list