Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1430996

Summary:	Write-behind doesn't pass the application credentials while syncing cached writes
Product:	[Community] GlusterFS	Reporter:	Raghavendra G <rgowdapp>
Component:	write-behind	Assignee:	Csaba Henk <csaba>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	unspecified	Docs Contact:
Priority:	medium
Version:	mainline	CC:	atumball, bugs, khiremat, vbellur
Target Milestone:	---	Keywords:	Triaged
Target Release:	---
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:	glusterfs-6.x	Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2019-07-02 04:12:21 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:
Bug Depends On:	1430608
Bug Blocks:

Description Raghavendra G 2017-03-10 05:13:43 UTC

Description of problem: It is not copying uid, pid, gid, ngroups, lk_owner, groups_large, groups, groups_small (various fields in call_stack/frame->root) etc during wb_fulfill. This is because wb_fulfill creates a new frame and a new frame created will have 0 set for all of these fields. This can result in various authentication mismatches by xlators like posix-acl etc. Note that these credentials have to be copied from frame->root to request as the original frame/call-stack might've been unwound and need not be alive at the time of wb_fulfill. One such issue can be seen in bz 1430608.

Another interesting fact to be looked at is that while collapsing multiple small writes into a single large holder, we should check for compatibility of these attributes across the requests that are collapsed.

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


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info: