[Bugs] [Bug 1403984] Node node high CPU - healing entries increasing

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1403984

Arie Skliarouk <skliarie+redhat-bugzilla at gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |skliarie+redhat-bugzilla at gm
                   |                            |ail.com



--- Comment #6 from Arie Skliarouk <skliarie+redhat-bugzilla at gmail.com> ---
For those who might have such problem, here is my situation and workaround:

On glusterfs 3.7.6-1ubuntu1 (ubuntu 16.04), we had had high CPU usage on one of
the gluster servers. It used all available CPU resources without writing
anything meaningful into logs. There were more than usual lines like these:

[2018-03-29 11:35:22.081500] I [MSGID: 109036]
[dht-common.c:7869:dht_log_new_layout_for_dir_selfheal] 2-gfs-cmj-dht: Setting
layout of /tmp/salt_www/dashman/cf37firsesc549t9a2a12o3653 with [Subvol_name:
gfs-cmj-replicate-0, Err: -1 , Start: 0 , Stop: 4294967295 , Hash: 1 ],  

Any file access to the glusterfs filesystem was slow, each operaton taking 10
seconds or more.

I found that this command mitigates the problem - high CPU went away and
glusterfs became fast again:
gluster volume gfs-cmj performance.readdir-ahead off

May this will help someone (took me many hours to find out).

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=rrhHvLL7sQ&a=cc_unsubscribe