Gluster users - Aug 2017 - How are bricks healed in Debian Jessie 3.11

Hi,

    How are bricks healed in Debian Jessie 3.11? Is it at the file of
block level? The scenario we have in mind is a 2 brick replica volume
for storing VM file systems in a self-service IaaS, e.g. OpenNebula. If
one of the bricks is off-line for a period of time all the VM files
systems will all have been modified when brick comes back on-line. As
some of these VM file systems are quite large, 100GB+, our guess is that
if healing involves copying entire files then it might be unworkable for
us as the time to restore a few minutes downtime of a single brick could
take days. Just copying modified blocks would be fine.

    Can someone give us some insight on how healing works?

            Regards,
                Gerry

-- 
Gerry O'Brien

Systems Manager
School of Computer Science and Statistics
Trinity College Dublin
Dublin 2
IRELAND

00 353 1 896 1341

On 08/08/2017 04:51 PM, Gerry O'Brien wrote:
> Hi,
>
>      How are bricks healed in Debian Jessie 3.11? Is it at the file of
> block level? The scenario we have in mind is a 2 brick replica volume
> for storing VM file systems in a self-service IaaS, e.g. OpenNebula. If
> one of the bricks is off-line for a period of time all the VM files
> systems will all have been modified when brick comes back on-line. As
> some of these VM file systems are quite large, 100GB+, our guess is that
> if healing involves copying entire files then it might be unworkable for
> us as the time to restore a few minutes downtime of a single brick could
> take days. Just copying modified blocks would be fine.
>
>      Can someone give us some insight on how healing works?
Healing of contents works at the entire file level at the moment. For VM 
image use cases, it is advised to enable sharding by virtue of which 
heals would be restricted to only the shards that were modified when the 
brick was down.
-Ravi
>
>              Regards,
>                  Gerry
>

> Healing of contents works at the entire file level at the moment. For VM 
> image use cases, it is advised to enable sharding by virtue of which 
> heals would be restricted to only the shards that were modified when the 
> brick was down.
We even change the heal algo to "full", since it seems better to just
re-download a small shard than trying to heal it. At least on 3.7 it
works better that way
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: Digital signature
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20170808/bdb7e062/attachment.sig>