[Gluster-users] High load on healing

[Edvin Ekström]

Hi,

I'm experiencing some issues with very long access times when the volume
is healing.

We're currently running a Gluster Distributed replicated 3x3 setup on
three servers which has been working flawlessly.
After adding a fourth server it was time for moving bricks around. The
problem we encountered was a big increase in accesstime on our gluster
while we were doing this healing work on the volume (from avg 0.052s to
avg around 6s).

Is there some way to configure gluster to use less capacity for healing/
Priorities external request or simply ionice the healing process?
From what I can tell the same PIDs were used for both healing and normal
requests so I have not been able to ionice it by external means.

Thank you in advance
Edvin Ekström

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0x44FEAAE9.asc
Type: application/pgp-keys
Size: 3116 bytes
Desc: not available
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20161220/05579686/attachment.bin>