Re: Status of MySQL on 7 w/o patches applied

From: Thomas Hurst <tom.hurst_at_clara.net>
Date: Tue, 16 Oct 2007 18:17:21 +0100
* Peter Schuller (peter.schuller_at_infidyne.com) wrote:

> My question is - what is the expected status of MySQL performance in
> FreeBSD 7 (without external patches) at this time?

Most if not all of the relevent patches were applied months ago.  We ran
MySQL in production for some time using both patched and later unpatched
7-CURRENT with ULE for a while, and normal performance was easily
comparible with Linux and Solaris.

We did see poor behavior with MySQL 5.1 and replication; the slave
thread needed periodic restarts because it would use progressively more
CPU and eventually fall behind, spending a long time in 'init' on every
query, which we didn't see in other OS's.  Whether this was FreeBSD's or
MySQL's fault I don't know, but one certainly didn't like the other.

> between 6.2 and 7-CURRENT. Selects were actually half as fast or less
> than 6.2 in this particular case (identical hardware).
> 
> Now I realize I am not providing details here; I can most definitely
> provide a lot more detail. But first, is this *expected*?

If you're running with WITNESS, INVARIANTS and/or haven't set
MALLOC_PRODUCTION, sure.  RELENG_7 has these disabled, but CURRENT
always tends to default to rather aggressive debugging options.

> Are there patches that still have not been committed, or should we be
> expecting that FreeBSD 7 as it appears in CVS is comparable to Linux
> or better with MySQL?

The only unapplied patches I'm aware of is Jeff's select locking patch,
which didn't seem to help MySQL very much:

http://lists.freebsd.org/pipermail/freebsd-arch/2007-August/006702.html

-- 
Thomas 'Freaky' Hurst
    http://hur.st/

Received on Tue Oct 16 2007 - 15:17:23 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:19 UTC