[Gluster-users] Tunning GlusterFS

Joe Julian joe at julianfamily.org
Fri Aug 2 17:45:52 UTC 2013


If I'm reading this right, you had a distribute volume and added replication, or had a disk full of files and created a replicated volume using that as a brick. If either of those is true:

Do "gluster volume heal $vol full" to make sure you're not just getting the new files and other files you touched.

If neither of those is true, then you have something else wrong. Your files should be replicated instantly as they're written. Make sure you are only using your volume through a client mount. Check your client logs for connection errors.

"Victor Hugo Martinucci Belizário" <vbelizario at tray.net.br> wrote:
>-- Hey Masters!
>
>Please help me.
>
>I'm testing GlusterFs on company, working with 2 milions of small
>files,
>total size its 200 GB.
>GlusterFs its a very powerfull toll to sync and replicate files, and i
> decide to try it.
>The files are in the server srv01, and i did create the server srv02 to
>replicate the files and initialize the sync between him.
>Created the volume, started the volume and the sync are ok.
>
>But...after 5 hours, just 2 GB of files are in the srv02. The sync its
>very
>very slow...less 1 MB/s.
>
>2 Questions:
>
>1) In the tunning options, what are the better options to tunning the
>volume to this cenario, a lot of small files with glusterFS ?
>
>2) i have read in some site, when a client read a file, the Gluster
>check
>all servers for the file integrity, and this make more slow the
>process.
>How i can turn off this feature?
>
>Thanks Very Much,
>
>Victor
>
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Gluster-users mailing list
>Gluster-users at gluster.org
>http://supercolony.gluster.org/mailman/listinfo/gluster-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20130802/5d63a885/attachment.html>


More information about the Gluster-users mailing list