Searching Tickets takes exponentially longer with each additional requestor

  • RT 5.0.2, with NGINX & MariaDB 10.5 (OR) PostgreSQL 13.7
  • Storage: qcow2, on gluster, on 10-disk zfs10 array.

We have approximately 100k tickets, so not a huge db, and recently upgraded from RT4 to RT5. We have run into a problem with some really long-running queries. Here’s an example:

Searching from the GUI for tickets from ‘requestor1’ takes less than 2 seconds. ‘requestor1’ OR ‘requestor2’ takes ~23 seconds. ‘requestor1’ OR ‘requestor2’ OR ‘requestor3’ takes 40-60+ seconds. Adding one more gets “Bad Gateway” from NGINX, and the Pg thread continues on indefinitely.

We exported/serialized our data, and reimported with rt-importer (tried both MySQL & Postgres) into a new-from-scratch DB, with the default indexes.

I’m not really good enough with DB analysis to look at an ‘explain’ statement for possible indexing needs, but is that where we need to go?

Or is it a storage problem? And if that…why the exponential increases?

Thanks for any suggestions!