[Gluster-users] help on AFR translator

[maurizio oggiano]

Hi,
I' m using AFR to keep aligned a file system on two servers.
This file system contain oracle data.
Oracle works properly on the file system managed by AFR either on the server
A or server B.
But if I switch off  server A while oracle is runnig and I launch oracle on
the second server all works fine.
When server A comes up again and i stop oracle on the server B and I try to
run oracle on the server A oracle doesn't work properly because there is
some file locked. ( After this operation oracle doesn't work on the server A
but works on the server B )
The only way to exit from this situation is to reboot server B, (the server
where oracle works properly).
Is there a way to force an unlock file or a way to avoid this behaviour?
Moreover when I switch off a server, the file system on the other server is
not available( the command ls remains blocked) for 25 seconds.
How can I do to have available the file system in a short time?


I'm using a 2.0.3 release and fuse 2.7.4.
I have two servers configured as follow:

client vol:
volume CRS
  type protocol/client
  option transport-type tcp/client
  option remote-host 127.0.0.1
  option remote-subvolume disk
end-volume

volume writeback-disk
  type performance/write-behind
  option block-size 131072 # unit in bytes
  subvolumes CRS
end-volume

volume readahead-disk
     type performance/read-ahead
     option page-size 65536     # unit in bytes
     option page-count 16       # cache per file  = (page-count x page-size)
     subvolumes writeback-disk
end-volume

server vol
   volume local-disk-ds
           type storage/posix
           option directory /var/glusterfs/shared
   end-volume

   volume TSU-1.localdomain-disk
       type features/locks
       subvolumes local-disk-ds
   end-volume

   volume remote.localdomain-disk
           type protocol/client
           option transport-type tcp/client
           option remote-host 192.168.1.74
           option remote-subvolume TSU-2.localdomain-disk
           option transport-timeout 5
   end-volume

  volume disk-afr
          type cluster/afr
          subvolumes remote.localdomain-disk TSU-1.localdomain-disk
          option favorite-child TSU-1.localdomain-disk
  end-volume

   volume disk-unify
           type cluster/distribute
           subvolumes disk-afr
   end-volume

   volume disk
           type performance/io-threads
           option thread-count 2
           subvolumes disk-unify
   end-volume

   volume server
     type protocol/server
     option transport-type tcp/server
     subvolumes disk
     option auth.addr.TSU-1.localdomain- disk.allow *
     option auth.addr.local-disk-ds.allow *
     option auth.addr.disk.allow *
   end-volume

Thanks Maurizio
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20090714/5cd48bf6/attachment.html>