[Gluster-users] Self-heal without reason

Miloš Kozák milos.kozak at lejmr.com
Wed Aug 26 06:28:51 UTC 2015


Hi,

I am trying to optimize my GlusterFS volume, the configuration is in 
attachement. My goal is to have glusterfs volume for VMs. This 
configuration seems to work and can deal with node failure which is what 
I really need.

So I after acceptance tests I have just moved to the performance test 
where I found an bug?


For my test, I have replica 2 volume which I fill up with VM images of 
different size. 10G file is totally OK, but 20G file is fine for whole 
time, but in the end load on uploading node increases significantly to 
20erl at 12threads i5 processor.. as long this is only for time it is 
OK. After that volume seems to be healing file for another 30minutes, so 
I can not carry on uploading images.

When I tried to upload 60G image, around 25G of uploaded data load got 
up to 30erl... then I had to stop the test because node was very 
unresponsive..

I am attaching logs because I can see there are some errors in brick 
log, but I can not say what the actual problems is, i.e., how to fix it..

Thanks, Milos
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gfs-s3-sata-10k-fs.log
Type: text/x-log
Size: 768513 bytes
Desc: not available
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150826/a88358bd/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: glustershd.log
Type: text/x-log
Size: 43440 bytes
Desc: not available
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150826/a88358bd/attachment-0003.bin>


More information about the Gluster-users mailing list