| 1 |
Chris Bainbridge wrote: |
| 2 |
> I'm curious what the problem is with bugzilla and it's db |
| 3 |
> interactions? You're suggesting a specific issue rather than general |
| 4 |
> db performance issues like fs, io scheduling, raid1, hyperthreads, |
| 5 |
> etc.? |
| 6 |
|
| 7 |
It's most likely related to Bugzilla using MyISAM tables by default |
| 8 |
which is fine in a small environment. However as concurrency grows along |
| 9 |
with table size those full table locks begin to cause issues. |
| 10 |
Additionally most www-apps are not known for their well thought out db |
| 11 |
schema. Performance of the underlying hardware plays a part, but can be |
| 12 |
overshadowed pretty quickly by query and table inefficiencies. |
| 13 |
|
| 14 |
The usual fix without touching the internals of the software is doing |
| 15 |
master/slave replication and allowing the selects to happen on the slave |
| 16 |
and writes on the master. However you would need to change most of the |
| 17 |
queries to point to the slave server which is usually not trivial. It |
| 18 |
sounds like this is the path the Infra team is pursuing. |
| 19 |
|
| 20 |
kashani |
| 21 |
-- |
| 22 |
gentoo-dev@g.o mailing list |
Archives