[Gluster-devel] strange thing with afr

[Albert Shih]

Hi All

I've very strange thing with afr.

I'm making a namespace with afr and it's not working....sometime.

First it's working well...I've reboot everything and ... it's not working.

Now I've re-initial everything in glusterfs (make new mke2fs etc...)
and...it's not working..

volume node1-ns
  type protocol/client
  option transport-type tcp/client     # for TCP/IP transport
  option remote-host 145.238.189.22
  option transport-timeout 30
  option remote-subvolume brick-ns
end-volume
volume node2-ns
  type protocol/client
  option transport-type tcp/client     # for TCP/IP transport
  option remote-host 145.238.189.23
  option transport-timeout 30
  option remote-subvolume brick-ns
end-volume
volume node3-ns
  type protocol/client
  option transport-type tcp/client     # for TCP/IP transport
  option remote-host 145.238.189.24
  option transport-timeout 30
  option remote-subvolume brick-ns
end-volume

# Definition du namespace via afr

volume ns
  type cluster/afr
  subvolumes node1-ns node2-ns node3-ns 
end-volume

I'm using in my unify volume 

 option namespace ns

And what's I mean when I say it's not working is when a ns create in
my unify volume the zero size file create is namespace is create only on
node1, not on node-2 or node-3.

Why that ? 

Other question about afr : It's that possible to re-synchronize the "raid"
? (for example when one node is off-line ? ) or it's automatic ? 

Regards.


--
Albert SHIH
Observatoire de Paris Meudon
SIO batiment 15
Heure local/Local time:
Mar 27 nov 2007 00:40:02 CET