Bug 763089 (GLUSTER-1357)

Summary:	Prospect reported issue with running multiple df commands simultaneously
Product:	[Community] GlusterFS	Reporter:	Jacob Shucart <jacob>
Component:	core	Assignee:	Anand Avati <aavati>
Status:	CLOSED WORKSFORME	QA Contact:
Severity:	medium	Docs Contact:
Priority:	low
Version:	3.0.4	CC:	chrisw, gluster-bugs, tejas
Target Milestone:	---
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Jacob Shucart 2010-08-13 21:24:02 UTC

They were unable to reproduce it, and they could not give me more details, but I thought I would put it here in case it happens again so we can see that it was reported.  If you would prefer I don't open bugs for this sort of thing please let me know:

OK, I'll try 3.0.5 and I'll tell you. As for the bug, I'm not able to reproduce it, but there's one more and quite serious bug that appears in
3.0.4 and it has appeared even in earlier version. It didn't appear in 3.0.5's changelog, so I think, that it's still in there.

It is very rare but it causes a lot of system "df" commands on all nodes and high load average. I think it happens when one node is extremely busy. After restarting this one problematic node all other nodes are fit again. I guess that "df-ing" is used for countig nodes' free space. I suggest there should be a limit for only one running "df" command, otherwise there are lots of them causing problems. The environment of this bug (configuration of nodes etc.) are similar as they were in my ticket #1734. Though, I'm not able to reproduce it.

Comment 1 Anand Avati 2010-10-01 05:10:56 UTC

The given description is insufficient to debug. df is extremely inexpensive and is unlikely to be the cause of high load averages. Without logs it is hard to debug. Also, given that there was a 'problematic node', it is likely that the cause of the issue is misbehaving hardware of that machine.