[Gluster-users] Pausing a brick

Franco Broi franco.broi at iongeo.com
Tue May 27 09:15:37 UTC 2014


On Tue, 2014-05-27 at 14:38 +0530, Vijay Bellur wrote: 
> On 05/26/2014 11:50 AM, Franco Broi wrote:
> > Hi
> >
> > Is there any way now, or plans to implement a pause or freeze function
> > for Gluster so that a brick can be taken offline momentarily from a DHT
> > volume without affecting all the clients?
> >
> 
> How long would your use case need the brick to be offline for? Would the 
> server hosting the brick also be offline when this happens?

I think up to 30 minutes wouldn't be unreasonable, just long enough to
remount the underlying filesystem. Pausing the whole server would also
work but letting the other bricks continue to run would be less
disruptive.

Could the gluster daemon close all open files and sit and wait for a
resume instruction?

> 
> -Vijay
> 





More information about the Gluster-users mailing list