Bug 762718 (GLUSTER-986)

Summary:	Race in internal locking for AFR transactions
Product:	[Community] GlusterFS	Reporter:	Pavan Vilas Sondur <pavan>
Component:	replicate	Assignee:	Pavan Vilas Sondur <pavan>
Status:	CLOSED DUPLICATE	QA Contact:
Severity:	high	Docs Contact:
Priority:	urgent
Version:	mainline	CC:	gluster-bugs, shehjart, tejas
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	---
Regression:	RTP	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Pavan Vilas Sondur 2010-06-04 11:05:07 UTC

The current framework of locking using internal locks is racy and can lead to corruption and self healing not working properly.

Apart from corruption, another problem can happen - self heal / flush (full file locks) can starve due to continuous I/O and never get a lock while I/O continues. So, sometimes a flush is not immediately observed when a child goes up/down. This is especially problematic since, self heal can be blocked when a child is up and I/O continues. This coupled with the race involved as mentioned above can lead to corruption.

See bug 762692 for more information.

Comment 1 Shehjar Tikoo 2010-07-27 08:40:51 UTC

Setting it duplicate of bug 762692, since thats where all the action has been in fixing this.

*** This bug has been marked as a duplicate of bug 960 ***