[Gluster-users] Painless upgrade?

Beat Rubischon beat at 0x1b.ch
Mon Mar 7 22:10:32 PST 2011


Hello!

I'm operating a GlusterFS installation running 2.0.7. After the good
experience with a patched 3.1.2 in an other project I think about an
upgrade.

It's a distribute/replicate setup over two servers with four bricks
each. Some of the files have metadata in their extended attributes:

	# getfattr -m "" -Rd .

	# file: directory
	trusted.afr.Data1VolA=0sAAAAAAAAAAAAAAAA
	trusted.afr.Data1VolB=0sAAAAAAAAAAAAAAAA
	trusted.glusterfs.dht=0sAAAAAQAAAAAAAAAAP////g==

	# file: directory/file
	trusted.afr.Data1VolA=0sAAAAAAAAAAAAAAAA
	trusted.afr.Data1VolB=0sAAAAAAAAAAAAAAAA

	# file: directory/anotherfile
	trusted.glusterfs.dht.linkto="Data2Mirror\000"

Is there any painless way to switch from the existing, manually
configured GlusterFS 2.x backend to the new glusterd generated setup in
3.1? An automatic or manual rewrite of the metadata?

Of course I'm able to recycle the existing volfiles and ignore glusterd,
but I like the new approach and I would prefer to use it. I'm also able
to dump / restore the content of the filesystem, but it will take ages
to do so :-)

Thanks for your feedback!
Beat

-- 
     \|/                           Beat Rubischon <beat at 0x1b.ch>
   ( 0-0 )                             http://www.0x1b.ch/~beat/
oOO--(_)--OOo---------------------------------------------------
Meine Erlebnisse, Gedanken und Traeume: http://www.0x1b.ch/blog/


More information about the Gluster-users mailing list