Hi folks,
Can someone give me an idea why gluster and the glusterd
would be taking almost 25% cpu each when under a READ load. (No writing).
I have setup just a simple three node system, and when I put it under a load
of a number of clients; gluster is really killing the cpu. Anything I
should specifically look at? Using v3.8.15.
Basically all three machines replicate each other; the idea is so that each
machine can be a member of the apache serving group for scalability. The
load test asked for basically the same 40 or so files over and over again.
If anyone has some pointers on what to look at; I can ramp up the load on
the machine on demand and test things; but having it take ~45% of the total
cpu under what I consider a fairly light load worries me a lot about
Glusters Scalability. Based on the bug reports and other newgroup
messages it seems I might be hitting some corner case as I believe this is
unusual result; so any pointers, any data I can provide.
Nathanael A.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.gluster.org/pipermail/gluster-users/attachments/20171204/9ff3ad99/attachment.html>
