Bug 1367881

Summary:	Replicate+arbiter volume size reported differently for NFS
Product:	[Community] GlusterFS	Reporter:	Brian Sipos <BSipos>
Component:	replicate	Assignee:	Ravishankar N <ravishankar>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	3.7.2	CC:	BSipos, bugs, pkarampu, ravishankar
Target Milestone:	---	Keywords:	Triaged
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2016-10-15 13:36:32 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Brian Sipos 2016-08-17 17:22:21 UTC

Description of problem:
I have created a replica-3 volume with one arbiter (having a much smaller disk than the other two). When mounted as glusterfs the volume shows correct total size (3.6TiB) but when mounted as NFS the volume shows the size of the arbiter brick (860GiB). I'm using this within oVirt, so the perceived volume size limits how much storage the oVirt engine will allow in a volume.

Version-Release number of selected component (if applicable):
3.7.2 built on Jun 23 2015 12:07:13

How reproducible:
always

Steps to Reproduce:
1. Create a volume with "gluster volume create testvol replica 3 arbiter 1 ..."
2. Mount the volume with "mount -tglusterfs ...:/testvol /mnt" and observe "df -h /mnt" shows large size. 
3. Unmount as glusterfs.
4. Mount the volume with "mount -tnfs4 ...:/testvol /mnt" and observe "df -h /mnt" shows much smaller size.
5. Unmount as NFS.

Actual results:
The "size" of the replicate volume is mis-reported as the size of the smallest brick, ignoring arbiter status.

Expected results:
The volume size should be computed from the non-arbiter bricks. At least the two network FS mounts should agree on how much space is available.

Additional info:

Comment 1 Brian Sipos 2016-08-17 18:29:22 UTC

I just realized that I am seeing this behavior from a client running on Fedora version "3.8.2 built on Aug 10 2016 15:29:39".

The servers are running the above listed version (3.7.2), and the servers actually see a consistent but incorrect view of "860GiB" volume size.

Comment 2 Ravishankar N 2016-08-30 14:23:37 UTC

(In reply to Brian Sipos from comment #1)
> I just realized that I am seeing this behavior from a client running on
> Fedora version "3.8.2 built on Aug 10 2016 15:29:39".
> 
> The servers are running the above listed version (3.7.2), and the servers
> actually see a consistent but incorrect view of "860GiB" volume size.

The fix (http://review.gluster.org/#/c/11857/) for this went in the 3.7.4 release. Can you upgrade your cluster to that version or later and  see if this is still happening?

It would be best if you can upgrade your entire setup to  the latest 3.8.x. version. There have been quite a few fixes for arbiter since 3.7.2.

Comment 3 Ravishankar N 2016-10-15 13:36:32 UTC

Further to comment #2, I'm closing the bug. Please re-open/ raise a bug if issue persists.

Comment 4 Red Hat Bugzilla 2023-09-14 03:29:43 UTC

The needinfo request[s] on this closed bug have been removed as they have been unresolved for 1000 days