Search Index Always Needs Manually Rebuilding

[heyjon]

No matter what we try our queue never optimizes itself anymore, so we always need to rebuild the search index manually. Support said our JVM heapsizes look fine based on our usage, and I've taken the steps to recreate the index from scratch. This fix seems to work for a week or so, but eventually we have to start rebuilding the search index everyday. Here is a screenshot taken on 6/27 (I've had to rebuild the search queue manually everyday since):

We're currently running Confluence 4.1.7 on a RHEL 6.2 VM server, with 4GBs of RAM. We may get a dozen updates per day. Any info that could lead us to a permanent solution would be highly appreciated.

[CharlesH]

Have you checked the scheduled jobs for flushing the index queue? Details here: https://confluence.atlassian.com/display/DOC/Scheduled+Jobs. According to the docs the default interval is 1 minute, so if this is the case and the job is failing you'll have some idea of what is going wrong thanks to the entries in the log file. What database are you running Confluence on?