Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1801615

Summary:	FUSE mount : touch exact the same file right after rm FAIL
Product:	[Community] GlusterFS	Reporter:	Cliff.Lu <Cliff.Lu>
Component:	fuse	Assignee:	Csaba Henk <csaba>
Status:	CLOSED NOTABUG	QA Contact:
Severity:	urgent	Docs Contact:
Priority:	unspecified
Version:	mainline	CC:	bugs, csaba, ravishankar
Target Milestone:	---	Keywords:	Triaged
Target Release:	---
Hardware:	Unspecified
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2020-03-02 03:03:57 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 Cliff.Lu 2020-02-11 11:24:28 UTC

Description of problem:
Touch exact the same file right after rm FAIL
We're using v4.1.8 and found this issue in v4.1.8.
But it could also be duplicate with master branch.

I've checked some log and code trace, 
It seems create/lookup fail before FUSE_FORGET for unlink comes.
I thought it shouldn't create/lookup as the same nodeid?

Version-Release number of selected component (if applicable):
tag/v4.1.8, branch/master

How reproducible:
It happens all the time if no other delays between sys call.
I've try 1000-times repeat, with 184 touch fail.
(The first time rm-touch almost 100% failed)

Steps to Reproduce:
1.
Create any type of volume(EC/Rep/Distribute)
mount with FUSE.
2.
rm aaa
touch aaa
with no other delay between two commands. (I've tried both shell, python)
3.
check touch fail

Actual results:
In master
touch: setting times of ‘/mnt/test/aaa’: Stale file handle
In v4.1.8
touch: setting times of ‘/mnt/test/aaa’: No such file or directory
file aaa not exist.

Expected results:
touch success
file aaa exist

Additional info:
OS - CentOS7.4

Comment 1 Ravishankar N 2020-02-18 07:09:12 UTC

> It seems create/lookup fail before FUSE_FORGET for unlink comes.

Csaba, would this be this something similar to the  template-xlator-sanity.t that you debugged recently?

Comment 2 Cliff.Lu 2020-03-02 03:03:57 UTC

Only pipeline cmd call trigger this one with FUSE.
I think it's not a bug.

Comment 3 Csaba Henk 2020-03-02 19:57:03 UTC

(In reply to Ravishankar N from comment #1)
> > It seems create/lookup fail before FUSE_FORGET for unlink comes.
> 
> Csaba, would this be this something similar to the  template-xlator-sanity.t
> that you debugged recently?

I don't think so, that phenomenon was specific to RHEL/Centos 8.