Bug 765235 (GLUSTER-3503)

Summary:	open should trigger data self-heal instead of lookup
Product:	[Community] GlusterFS	Reporter:	Pranith Kumar K <pkarampu>
Component:	replicate	Assignee:	Pranith Kumar K <pkarampu>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	medium	Docs Contact:
Priority:	medium
Version:	mainline	CC:	gluster-bugs, rabhat
Target Milestone:	---
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:		Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:	glusterfs-3.3.0qa10	Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Pranith Kumar K 2011-09-02 03:40:14 UTC

In afr, data self heal will happen only in open or fd based fops but can perform it in lookup itself. This change is brought in number of back-ground self-heals allowed is 16 and lookups block self-heal is completed. We want to prevent blocking in fops. We can not lookups independent of self-heal frames because when there are gfid the decision of which file is correct is determined in self-heal So in afr, lookup self-heal is going to guarantee name space and open/fd fops will take responsibility for data consistency, are non blocking.

Comment 1 Anand Avati 2011-09-06 06:15:59 UTC

CHANGE: http://review.gluster.com/333 (This is brought in an effort to be nice to the system resources when) merged in master by Anand Avati (avati)

Comment 2 Anand Avati 2011-09-09 00:33:57 UTC

CHANGE: http://review.gluster.com/334 (By default, lookup triggers data self-heal but that is not the preferred way) merged in master by Anand Avati (avati)

Comment 3 Raghavendra Bhat 2011-09-23 06:07:51 UTC

Its fixed now. Checked with glusterfs-3.3.0qa10. Created a replicate volume, and did this.

gluster volume set <volname> data-self-heal open
gluster volume set <volname> self-heal-daemon off

mounted the volume and touched a file. Brought one of the servers down, wrote some data to the file. Bought the server up. Doing ls or stat on the file did not heal it. When I did cat on the file, it's data got healed.