Bug 764822 (GLUSTER-3090)

Summary:	data self-heal is not happening with 3.2.1 rdma set-up
Product:	[Community] GlusterFS	Reporter:	M S Vishwanath Bhat <vbhat>
Component:	replicate	Assignee:	Raghavendra G <raghavendra>
Status:	CLOSED NOTABUG	QA Contact:
Severity:	medium	Docs Contact:
Priority:	medium
Version:	3.2.1	CC:	amarts, gluster-bugs, mzywusko, pkarampu
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:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description M S Vishwanath Bhat 2011-06-27 11:24:11 UTC

I created a 2 way replicate volume. I created 1 million 10K files on the mount point through dd command. When the creation of files were half way through (just around 5000 files) I brought the first replica child down. Now after the creation of 1 million files, I brought up the server which was down. Now I run 'find .' command on the mount point to trigger the self-heal. But after a while I see that only entries are self healed (There was 1 million files on the brick). But size was just 70MB around while the original data is around 16GB.

I was able to hit the problem twice. The transport type was 'rdma'. 

The logs were too big to even read because I had et the log level to DEBUG. I will try to reproduce this with lesser data.

Comment 1 Pranith Kumar K 2011-07-01 06:18:35 UTC

(In reply to comment #0)
> I created a 2 way replicate volume. I created 1 million 10K files on the mount
> point through dd command. When the creation of files were half way through
> (just around 5000 files) I brought the first replica child down. Now after the
> creation of 1 million files, I brought up the server which was down. Now I run
> 'find .' command on the mount point to trigger the self-heal. But after a while
> I see that only entries are self healed (There was 1 million files on the
> brick). But size was just 70MB around while the original data is around 16GB.
> 
> I was able to hit the problem twice. The transport type was 'rdma'. 
> 
> The logs were too big to even read because I had et the log level to DEBUG. I
> will try to reproduce this with lesser data.

Vishwa tried the same test with 50,000 files and the same happened but when the command "find <mount> | xargs ls -l" is executed on the mount point, everything healed fine, please verify that the self-heal is triggered using the command "find <gluster-mount> -noleaf -print0 | xargs --null stat >/dev/null".

Comment 2 Raghavendra G 2011-07-11 01:29:12 UTC

MS/Pranith,

Can we close this bug as invalid (since wrong command was used for self-healing)?

regards,
Raghavendra.