[Gluster-users] Heal rate throttle

Michael Mol mikemol at gmail.com
Fri Jul 17 21:32:08 UTC 2015


So, today, I needed to reboot one of my bricks to clear out a file handle
to a deleted file that glusterfsd was holding on to. During the heal
process, my infrastructure has slowed to a crawl. It's not something I
could tolerate during normal business hours.

With any plain old raid array, I can tune the priority the raid controller
gives to rebuilding the array versus what it gives to I/O consumers. But
I'm not seeing anything like that in gluster. There's
cluster.data-self-heal-algorithm,
and there's cluster.self-heal-window-size, but it's not clear how they'd be
useful in a volume holding large disk images, and they certainly don't seem
to represent the idea of a tunable priority.

What am I missing? Am I just looking at this the wrong way?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150717/e03e7826/attachment.html>


More information about the Gluster-users mailing list