[Bugs] [Bug 1657743] Very high memory usage (25GB) on Gluster FUSE mountpoint

bugzilla at redhat.com bugzilla at redhat.com
Thu Jul 11 05:16:35 UTC 2019


--- Comment #18 from Nithya Balachandran <nbalacha at redhat.com> ---
Hi Ryan,

As the volume seems to have changed from that in the description (3 bricks
instead of 15), I think it best we start over. Can you get us the following:

1. Statedump of the client process immediately after the volume is mounted
(this will be the baseline)
2. The statedumps at intervals when the client is being used.
3. Is there any point at which the client mount will be idle (no operations
happening)? If yes, a statedump at that time as well to get a clearer picture
of the memory usage when there are no fops in progress.

The earlier statedumps indicated an increasing number of gf_common_mt_memdup
allocations which is what I thought might be contributing to the leak. However,
they could also mean that the client was doing a lot of file listings at the
time. If this is the leak, the two patches posted in an earlier comment should
fix them.

As you also have client-io-threads enabled, another possibility is that you
might be running into https://review.gluster.org/#/c/glusterfs/+/22918.

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

More information about the Bugs mailing list