[Gluster-users] Trasnference speed slow.

[Curro Rodriguez]

Hello,

I have a replica with one server with 12TB of data used and new one added
later, I am syncing them, started using

gluster volume heal datastore full

To trigger the healing, but It was slow. After I was looking for a better
solution, I saw in previous mails on this list that I could sync faster
using stat on mounted clients.

So I tried

find -exec stat {} \; > /dev/null

Improved a lot the speed, but still seems slow to me, I have around 5MB/s
getting close to 20GB per our of transference.

I know that the reason of low speed could be a lot of small files, but I
would like to know if is there anything more to improve this speed.

Machines are 1 Gbps Bandwidth (node-1) to 10Gbps Bandwidth (node-2).
Gluster version 3.7.2 on Centos 7.1

I increased:
performance cache size to 1GB
performance chache max file size to 256KB
performance io-thread-count to 32

If you need more info about stats, machines, etc please ask me.

Thank you for your help.

Regards.

Curro Rodríguez
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150708/64e5f23c/attachment.html>