I’m trying to optimize our RT installation performancewise a bit. It has
lately started to grow slower and slower which is kind of annoying. I’ve
tried to locate the culprit, and it feels like it is something in the
web/perl/HTML::Mason part of it that is causing the slowdown since the
MySQL server is nowhere near having high load (CPU nor I/O) figures.
[I also tried enabling the MySQL “slow-queries” log - and it has
turned up zilch].
Hardware:
Sun Netra T1 with a 500MHz CPU, 1GB RAM and mirrored 36GB
Seagate Cheetah 10000rpm disks - dedicated to the MySQL server
for the RT system (currently). Solaris 8.
Sun Netra X1 with a 500MHz CPU, 1GB RAM and mirrored 40GB
Seagate Barracuda IV 7200rpm IDE disks - dedicated to the
Apache/mod_perl/HTML::Mason RT system. Solaris 8.
The servers talk to each other over a 100Mbps ethernet link.
The RT installation is nearing ticket #10000 now, 855 users
(according to “rt --list-users|wc”) - of which about 10 are
privileged (the rest are the autocreated sort) and about 10 queues
(of which one is used a lot and two somewhat).
It typically takes 5-10 seconds for the system to present a view
of tickets, or 2-5 seconds when updating something in a ticket
(like killing it…).
When browsing the archives I found a notice of missing SQL indexes,
but as far as I can see all the primary indexes are there as
they should be. (And that problem was indicated by the MySQL
server going to 100% CPU)
Another suggestion talked about ACL evaluations problems in RT2
that was fixed in RT3 - can this be what I’m seeing here?
Peter Eriksson peter@ifm.liu.se Phone: +46 13 28 2786
Computer Systems Manager/BOFH Cell/GSM: +46 705 18 2786
Physics Department, Link�ping University Room: Building F, F203
SE-581 83 Link�ping, Sweden http://www.ifm.liu.se/~peter