[Gluster-users] Understanding gluster performance

Yaniv Kaul ykaul at redhat.com
Tue Jan 21 10:40:17 UTC 2020


On Tue, Jan 21, 2020 at 11:55 AM Gionatan Danti <g.danti at assyoma.it> wrote:

> Il 20-01-2020 16:30 Gionatan Danti ha scritto:
> > So, I have some questions:
> > - why performance are so low with fsync?
> > - why do I have so low IOPs (20/30) for minutes?
> > - what is capping the non-fsync test?
> > - why both glusterd and glusterfd are so CPU intensive? I can
> > understand glusterfd itself requiring more CPU, but glusterd should
> > only manage the other processes and send the volume information to
> > asking client, right?
>
> An update: the very low IOPs were caused by TCP retransmission on one of
> the two boxes. After that problem was fixed, no such low performance was
> recorded.
>

How did you fix this?
How did you spot this?

>
> I also think that the non-fsync tests, with about 5000 IOPs, are cappaed
> by the 1000BaseT round trip time.
>
> Some questions remain:
> - why performance are so low with fsync? (about 250 IOPs)
>

fsync requires each write to land on disk and get an ack on it - it's
probably the slowest kind of write you can imagine, and you seem be doing
it for every small (4K?) block.
This is not very realistic. But in your case you say you are writing to
/dev/shm, so that is strange. Can you try with a different fio engine and
see what you get?
What are you trying to test here?


- why both glusterd and glusterfd are so CPU intensive?
>

Good question. Perhaps profiling would help here. Perhaps too many threads
are contending for CPU? Some lock contention?
Y.

>
> Thanks.
>
> --
> Danti Gionatan
> Supporto Tecnico
> Assyoma S.r.l. - www.assyoma.it
> email: g.danti at assyoma.it - info at assyoma.it
> GPG public key ID: FF5F32A8
> ________
>
> Community Meeting Calendar:
>
> APAC Schedule -
> Every 2nd and 4th Tuesday at 11:30 AM IST
> Bridge: https://bluejeans.com/441850968
>
> NA/EMEA Schedule -
> Every 1st and 3rd Tuesday at 01:00 PM EDT
> Bridge: https://bluejeans.com/441850968
>
> Gluster-users mailing list
> Gluster-users at gluster.org
> https://lists.gluster.org/mailman/listinfo/gluster-users
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gluster.org/pipermail/gluster-users/attachments/20200121/3ea2d403/attachment.html>


More information about the Gluster-users mailing list