[Bugs] [Bug 1623107] FUSE client's memory leak

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1623107

Travers Carter <tcarter at noggin.com.au> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tcarter at noggin.com.au



--- Comment #34 from Travers Carter <tcarter at noggin.com.au> ---
We are seeing what looks like the same issue on glusterfs 4.1.5

I'm not sure if further information is still needed, given the last few
comments, but I've collected client statedumps from three systems, along the
with the gluster volume info here:

https://s3.amazonaws.com/public-rhbz/glusterfs-client-4.1.5-statedumps.zip

This includes 2 x client statedumps from each of 3 systems taken roughly 30 to
60 minutes apart

The "webserver" and "appserver-active" gluster clients were restarted after
setting readdir-ahead to off as suggested earlier in the ticket (this didn't
seem to help much in this case), but the "webserver" client has already reached
about 15GiB VIRT in just over 48 hours.

We had also historically seen somewhat slower, but still significant fuse
client memory leaks on v3.x (I think 3.11 or 3.12), but not (or at least not
significant) on 3.7.11 or 4.0.2 with very similar workloads.

-- 
You are receiving this mail because:
You are on the CC list for the bug.