Bug 1401473

Summary:	occassional stale fuse mounts exist even after unmount of fuse volume
Product:	Red Hat Gluster Storage	Reporter:	Nag Pavan Chilakam <nchilaka>
Component:	fuse	Assignee:	Ravishankar N <ravishankar>
Status:	CLOSED WORKSFORME	QA Contact:	Rahul Hinduja <rhinduja>
Severity:	medium	Docs Contact:
Priority:	low
Version:	rhgs-3.2	CC:	atumball, csaba, nchilaka, rhs-bugs, storage-qa-internal
Target Milestone:	---	Keywords:	ZStream
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2018-11-26 14:18:47 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:

Description Nag Pavan Chilakam 2016-12-05 11:45:38 UTC

Description of problem:
=======================
I have observed this inconsistently but quite occassionally that even after I umount the fuse mount even with using -l (forceful) option, I see the fuse mount process still existing
For eg  I unmounted /mnt/rep2 
but ps -ef|grep rep2 shows me as below

root     32337     1  0 Dec02 ?        00:24:02 /usr/sbin/glusterfs --volfile-server=10.70.35.196 --volfile-id=rep2 /mnt/rep2

I did try check the fuse log but didnt find anything helpful

Version-Release number of selected component (if applicable):
============
3.8.4-6

This can lead  to memory consumption unwantedly

[root@rhs-client45 rep2]# top -n 1 -b|grep 32337
32337 root      20   0 4775952 3.760g   4348 S   0.0 24.3  24:02.77 glusterfs

Comment 2 Nag Pavan Chilakam 2016-12-05 12:20:21 UTC

client statedump post umount with stale process seen
[qe@rhsqe-repo bug.1401473]$ pwd
/var/www/html/sosreports/nchilaka/bug.1401473

Comment 3 Ravishankar N 2016-12-06 05:10:52 UTC

Nag, the -l option is to perform a lazy umount. There is no force option for fuse mounts. Do you see the "Unmounting <fuse mount point" message in the mount log (I'm not sure how to access the logs in comment #2)? The umount will happen only after all programs accessing the mount stop and you 'cd` out of the mount point.

Comment 4 Nag Pavan Chilakam 2016-12-06 12:56:56 UTC

We are deferring this to 3.2.0_beyond as we don't see visible impact other than resource consumption by the stale process

Comment 5 Ravishankar N 2016-12-06 14:18:43 UTC

(In reply to nchilaka from comment #4)
> We are deferring this to 3.2.0_beyond as we don't see visible impact other
> than resource consumption by the stale process

That seems reasonable Nag. But nevertheless, please attach the mount log and the statedump to the BZ as requested for in comment #3.

Comment 7 Nag Pavan Chilakam 2017-08-30 05:48:37 UTC

(In reply to Ravishankar N from comment #3)
> Nag, the -l option is to perform a lazy umount. There is no force option for
> fuse mounts. Do you see the "Unmounting <fuse mount point" message in the
> mount log (I'm not sure how to access the logs in comment #2)? The umount
> will happen only after all programs accessing the mount stop and you 'cd`
> out of the mount point.

I did cd out from all terminals. I did see the issue then.

Comment 9 Amar Tumballi 2018-11-26 09:31:36 UTC

Can this be tested with later versions, and verified? We have not hit the issues in long time now.