Bug 764512 (GLUSTER-2780)

Summary:	geo-replication operations take too much time to complete
Product:	[Community] GlusterFS	Reporter:	Csaba Henk <csaba>
Component:	geo-replication	Assignee:	Csaba Henk <csaba>
Status:	CLOSED CURRENTRELEASE	QA Contact:
Severity:	low	Docs Contact:
Priority:	medium
Version:	mainline	CC:	aavati, gluster-bugs, lakshmipathi
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:	RTNR	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:

Description Csaba Henk 2011-04-15 21:37:50 UTC

I'm using a system with no peers and volumes of single brick, so we'd expect no communication overhead.

gluster volume geo-replication status -- 3.2 sec: tolerable.
gluster volume geo-replication start  -- 9.5 sec: hmmmm.
gluster volume geo-replication stop   -- 26.6 sec: eeeeeeeeeeeeeeeeeeeeeeeek. 

Note that it's only on my system, on aws we do get ~3 sec runtimes for each operation. As otherwise my system performs well, still an interesting question why so.

Comment 1 Csaba Henk 2011-04-15 23:52:41 UTC

Comment 2 Anand Avati 2011-04-16 14:51:32 UTC

PATCH: http://patches.gluster.com/patch/6922 in master (syncdaemon: load xattrs from libc on-demand)

Comment 3 Anand Avati 2011-04-16 14:51:37 UTC

PATCH: http://patches.gluster.com/patch/6923 in master (glusterd: refactor gsync_status() so that we can get at the pidfile)

Comment 4 Anand Avati 2011-04-17 11:39:08 UTC

PATCH: http://patches.gluster.com/patch/6924 in master (glusterd: some cleanups needed for 70adbe7b [refactor gsync_status() ...])

Comment 5 Lakshmipathi G 2011-04-18 04:28:30 UTC

tested with 3.2.0qa14 on aws.
---
# time gluster volume geo-replication stop beta1 root.compute.amazonaws.com::slave 
geo-replication session stopped successfully

real	0m1.626s
user	0m0.007s
sys	0m0.010s
[root@ip-10-170-205-102 mntpt]# time gluster volume geo-replication start beta1 root.compute.amazonaws.com::slave 
geo-replication session started Successfully

real	0m1.061s
user	0m0.005s
sys	0m0.011s
-----

Comment 6 Anand Avati 2011-04-18 04:35:15 UTC

> geo-replication session stopped successfully
...
> geo-replication session started Successfully

Can someone fix the cases? This looks very unpolished.

Avati

Comment 7 Csaba Henk 2011-04-18 04:55:02 UTC

(In reply to comment #6)
> > geo-replication session stopped successfully
> ...
> > geo-replication session started Successfully
> 
> Can someone fix the cases? This looks very unpolished.
> 
> Avati

Kaushik has already taken that up.