[Gluster-users] gluster taking a lot of CPU and crashes database at times
joe at julianfamily.org
Wed Aug 8 16:54:38 UTC 2012
> On Wed, Aug 8, 2012 at 4:30 PM, Philip Poten <philip.poten at gmail.com
> <mailto:philip.poten at gmail.com>> wrote:
> running postgres (or any database) on a gluster share is an extremely
> bad idea. This can and will not end well, no matter what you do.
I still don't see why people keep saying this. I've been running mysql
on a GlusterFS volume since the 2.0 days. I know Avati agrees with you
though (though I keep trying to convince him otherwise).
The only problem I've ever had was with creation or alteration of MyISAM
files as they create a temporary filename, then rename it. This often
causes an error as the rename (apparently) hasn't completed before it
tries to open again (a bug that still seems to exist in 3.3.0).
InnoDB files can actually be quite efficient on a distributed volume if
you create sufficient file segments to be distributed across subvolumes.
The only real problem would be if someone thought they could run
multiple instances of the database server. Regardless of what filesystem
they're on, relational database engines are not built to be able to do that.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Gluster-users