[Bugs] [Bug 1767264] glusterfs client process coredump

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1767264



--- Comment #13 from Arie Skliarouk <skliarie+redhat-bugzilla at gmail.com> ---
Created attachment 1645626
  --> https://bugzilla.redhat.com/attachment.cgi?id=1645626&action=edit
glusterfs 7 client crash logs

Not sure whether my problem has same roots as the one under discussion.
Here is our gluster setup. We don't need speed, but we need HA.
Bricks on gl1 and gl2 sit on SSD, bricks on gl3 machine sit on magnetic
storage. The idea is that gl3 is used mostly as backup (1 full copy) while gl1
and gl2 keep half copy each (using the smallish arbiter brick).

Volume Name: nafsha
Type: Distributed-Replicate
Volume ID: fdbf91e6-0607-46c8-b538-103180b5ab19
Status: Started
Snapshot Count: 0
Number of Bricks: 2 x (2 + 1) = 6
Transport-type: tcp
Bricks:
Brick1: gl1:/mnt/nafshab1/BRICK
Brick2: gl3:/mnt/nafshab1/BRICK
Brick3: gl2:/mnt/nafshab1/BRICK (arbiter)
Brick4: gl2:/mnt/nafshab2/BRICK
Brick5: gl3:/mnt/nafshab2/BRICK
Brick6: gl1:/mnt/nafshab2/BRICK (arbiter)
Options Reconfigured:
transport.address-family: inet
storage.fips-mode-rchecksum: on
nfs.disable: on
performance.client-io-threads: off

The problem is that when we start rsync of many small files onto glusterfs
client (on a 4th machine), the client reproduceably crashes, after several
minutes of operation. See the attached file with details of the different
resulting crashes.

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