Hi kinglong,
I did not hit the same issue on my setup.. 4 cores + 16 GB ram.
I am planning to run some test by limiting no of cores and memory available with cgroups.. I did not hit the same issue on my setup.. 4 cores + 16 GB ram.
maybe there is an issue when the system has lower end configurations.
On Thu, Jun 29, 2017 at 6:21 PM, Sanoj Unnikrishnan <sunnikri@xxxxxxxxxx> wrote:
SanojRegards,Hi kinglong,
I will do this test and get back to u tomorrow.On Wed, Jun 28, 2017 at 6:23 PM, Kinglong Mee <kinglongmee@xxxxxxxxx> wrote:I use glusterfs quota on Centos7 with glusterfs-3.10.2-1.el7.x86_64, system memory is 2GB.
My test environment is two glustefs nodes, with one 10GB disk (ext4) each.
When enable quota with many files in the glusterfs, the glusterfsd in one node
costs high memory towards 40%+ of 2GB, or both nodes costs high memory sometimes.
Node one,
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
30017 root 20 0 1487412 115528 4648 S 0.0 6.2 3:11.81 glusterfsd
The other node,
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
102510 root 20 0 1439044 700264 4668 S 0.0 37.5 2:31.37 glusterfsd
The reproduce is,
# gluster volume start fsname
# mount -t glusterfs nodename:fsname /mnt/test/
# cp -rf linux-source /mnt/test/ <---- it's important adds many files before enable quota.
# gluster volume quota fsname enable
Sometime, needs disable quota and re-enable again, the high memory cost will appear.
But, I can't find the high memory cost when debugging by valgrind,
the memory cost will increase slowly to 10%.
Any comments are welcome.
thanks,
Kinglong Mee
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxxx
http://lists.gluster.org/mailman/listinfo/gluster-devel
_______________________________________________ Gluster-devel mailing list Gluster-devel@xxxxxxxxxxx http://lists.gluster.org/mailman/listinfo/gluster-devel
