[Gluster-devel] Disperse volume performance improvement using clients-io-theads

[jayakrishnan mm]

Hi,
Glusterfs Ver: 3.7.6

1) Currently I am using disperse  volume  and its performance is not
satisfactory.
 So I tried  "sudo gluster v set ec-vol performance.client-io-threads on"
(as per https://bugzilla.redhat.com/show_bug.cgi?id=1349953)

on the volume . But  I am getting
Connection failed. Please check if gluster daemon is operational.  error.

volume info

Volume Name: ec-vol
Type: Disperse
Volume ID: 66d9f9e9-8bbd-46d8-b491-5b3e48219ee1
Status: Started
Number of Bricks: 1 x (4 + 2) = 6
Transport-type: tcp
Bricks:
Brick1: jk:/home/jk/gluster/brick1
Brick2: jk:/home/jk/gluster/brick2
Brick3: jk:/home/jk/gluster/brick3
Brick4: jk:/home/jk/gluster/brick4
Brick5: jk:/home/jk/gluster/brick5
Brick6: jk:/home/jk/gluster/brick6
Options Reconfigured:
performance.readdir-ahead: on


Attached  the  log.

How  to resolve this error ?

2) If I replace  the Erasure code algorithm  in the EC  with a simpler one
, will  it  give a better performance ?  What is  the real bottleneck ?


Best Regards
JK
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160822/d861cbc9/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: usr-local-etc-glusterfs-glusterd.vol.log-3
Type: application/octet-stream
Size: 8520 bytes
Desc: not available
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160822/d861cbc9/attachment.obj>