[Gluster-devel] Question on Geo-Replication

Vijaykumar Koppad vkoppad at redhat.com
Mon Jul 23 18:49:58 UTC 2012


Hi Pitichai,

   First of all , having AFR over WAN is not a favorable setup. 
Things might go wrong with that setup. And Geo-rep over LAN is also 
not like using it efficiently. I can't say right now , why those files 
didn't get synced , unless i look into the logs. And there is nothing much 
about pre-existing data with Geo-rep.
     And about erasing indexing , you should do gluster volume set <vol-name> indexing off,
after stopping the geo-rep session. When you start geo-rep again , it will assign 
xtime to all the files which doesn't have xtime and starts syncing.   
 
Thanks,
Vijaykumar

----- Original Message -----
From: "Pitichai Pitimaneeyakul" <pitichai at 2-cans.com>
To: "Vijaykumar Koppad" <vkoppad at redhat.com>
Cc: "gluster-devel" <gluster-devel at nongnu.org>
Sent: Monday, July 23, 2012 6:09:36 PM
Subject: Re: [Gluster-devel] Question on Geo-Replication

Hi Vijaykumar, 
Thank you for quick response. :) 

This is what happen in Our Lab. We are setting as below with Gluster 3.3.0 on Debian 

Server-C (company location) <=== AFR over WAN ===> { Server-A(ISP) ==Geo Replication over LAN ==> Server-B(ISP) } 

I have configure geo replication for Volume (Vol01) which is created by Server-C brick and Server-A brick as AFR to A volume (Vol02) from Server-B's brick. 
The Vol01 have some data in side. 

After I ran "gluster volume geo-replication Vol01 serverB:Vol02 start", It starts replicating but some data under sub directories did not get replicated. 
I have the data set look like this 
data (Vol01) 
|-- atest01.txt 
|-- atest01.mp3 
|-- home 
| `-- users 
| |-- index.html 
| `-- somchai 
| |-- files 
| | `-- test.mp3 
|-- test01.txt 
|-- test01.mp3 

On the replication is look like 

Vol02 
|-- atest01.txt 
|-- atest01.mp3 
|-- home 
| `-- users 
|-- test01.txt 
|-- test01.mp3 

Not all the data and directory structure under /data are replicated, everything look OK except the data under /data/home/users/ is not appear. Even I try to copy some new file into /home/user/somchai directory. 
However, I backed up data and recreated Vol01, Vol02 volume with same configure of replication. I copied data from backup to Vol01, and the replication is correct and complete on Vol02 with full directory tree. If there any thing about preexisting data with geo-replication ? 

As Gluster troubleshooting guide said how to enforce full geo-replication by erasing index (page 115, section 14.2.3), and restart geo-replication but I don't see anything mention about erasing indexing in "Tuning Volume Options" as the document said. I would like to confirm the operation of erasing index is "set geo-replication.index=off" or remove ".glusterfs" under Vol01 and Vol02. 

For the log file, I will see I can retrieve it or someone already delete them. 

Thank you and Best Regards, 
Pitichai 





On 07/23/2012 03:05 PM, Vijaykumar Koppad wrote: 


Hi Pitichai,

    It will be helpful to find out the problem, if you provide us 
with the log files and the configuration of the setup. 
And also the data set you are using , I mean the size and number of  files. 

Thanks,
Vijaykumar 

----- Original Message -----
From: "Pitichai Pitimaneeyakul" <pitichai at 2-cans.com> To: "gluster-devel" <gluster-devel at nongnu.org> Sent: Saturday, July 21, 2012 1:53:24 PM
Subject: [Gluster-devel] Question on Geo-Replication




Hi there, 


I ran into problem of synchronization in geo-replication is not completed. 
And I found the the document say about force to full sync by erasing index and restart geo-replication 
as below from Gluster Document. Somehow I went to Tuning Volume Options, there is nothing mention about erasing index. 
Or it is just to set geo-replication.index=off and restart geo-replication ? 


====== 


Synchronization is not complete 

Description: GlusterFS Geo-replication did not synchronize the data completely but still the geo-replication status display OK. 

Solution: You can enforce a full sync of the data by erasing the index and restarting GlusterFS Geo-replication. After restarting, GlusterFS Geo-replication begins synchronizing all the data, that is, all files will be compared with by means of being checksummed, which can be a lengthy /resource high utilization operation, mainly on large data sets (however, actual data loss will not occur). If the error situation persists, contact Gluster Support. 

For more information about erasing index, see Tuning Volume Options . 

====== 

Thank you and Best Regards, 
Pitichai 

_______________________________________________
Gluster-devel mailing list Gluster-devel at nongnu.org https://lists.nongnu.org/mailman/listinfo/gluster-devel 

-- 




More information about the Gluster-devel mailing list