nm, I understand now what you mean by the empty export directory. I'll
try and reconcile the file differences with rsync and start with a fresh
export directory.
Thanks,
Erik
Erik Osterman wrote:
Yikes, I didn't realize that files must be unique per subvolume of a
unify translator. Guess I need to gc my glusterfs now. Hrm...
Any suggestions?
Is this a feature to be addressed in 1.4 with the f* stuff?
Thanks,
Erik
Amar S. Tumballi wrote:
Erik,
Yes, AFR replicates all the files (symlinks too) on its child nodes,
but UNIFY requires the file (not directories) on only one of the
child node.
Thus, as of now (with this release) when one is changing
configuration related to xlators/cluster/{afr,stripe,unify} the empty
export directory is recommended to avoid inconsistancy.
-amar
(bulde on #gluster)
On Fri, May 04, 2007 at 01:10:33AM -0700, Erik Osterman wrote:
Yes, the configuration prior to this had only 1 AFR layer with all 4
volumes inside of it. I had a problem getting uniform distribution
of files between the glusterfsd servers and got some help from bulde
on #gluster who showed me how to do it correctly.
Is that the cause?
Erik
Anand Avati wrote:
were the files in this created *after* coming to this configuration?
or were the files already created with a different configuration and
later on reconfigured?
avati
On 5/4/07, Erik Osterman <e@xxxxxxxxxxxx> wrote:
### Server
volume ourcompany
type storage/posix
option directory /home/glusterfs/ourcompany.com
end-volume
volume server
type protocol/server
subvolumes ourcompany
option transport-type tcp/server
option bind-address 0.0.0.0
option listen-port 6996
option client-volume-filename /etc/glusterfs/ourcompany-client.vol
option auth.ip.ourcompany.allow *
end-volume
### Client
volume ip0
type protocol/client
option transport-type tcp/client
option remote-host 1.2.3.23
option remote-port 6996
option remote-subvolume ourcompany
end-volume
volume ip1
type protocol/client
option transport-type tcp/client
option remote-host 1.2.3.24
option remote-port 6996
option remote-subvolume ourcompany
end-volume
volume ip2
type protocol/client
option transport-type tcp/client
option remote-host 1.2.3.25
option remote-port 6996
option remote-subvolume ourcompany
end-volume
volume ip3
type protocol/client
option transport-type tcp/client
option remote-host 1.2.3.26
option remote-port 6996
option remote-subvolume ourcompany
end-volume
volume afr1
type cluster/afr
subvolumes ip0 ip1
option replicate *:2 # All files 2 copies (RAID-1)
end-volume
volume afr2
type cluster/afr
subvolumes ip2 ip3
option replicate *:2 # All files 2 copies (RAID-1)
end-volume
volume ip
type cluster/unify
subvolumes afr1 afr2
option scheduler rr
option rr.limits.min-free-disk 2GB
end-volume
Anand Avati wrote:
can you please post your client and server spec file too?
avati
On 5/4/07, Erik Osterman <e@xxxxxxxxxxxx> wrote:
We are running the latest TLA as of around May 2, 3pm.
lrwxrwxrwx 1 daemon ourproduct 12 2007-04-30 13:43
shared ->
/mnt/shared/
lrwxrwxrwx 1 daemon ourproduct 12 2007-04-30 13:43
shared ->
/mnt/shared/
lrwxrwxrwx 1 ourproduct ourproduct 1 2007-04-30 13:43
ourproduct
-> .
lrwxrwxrwx 1 ourproduct ourproduct 1 2007-04-30 13:43
ourproduct
-> .
drwx------ 2 ourproduct ourproduct 4096 2006-10-24 23:24 .ssh
drwxr-xr-x 3 ourproduct ourproduct 4096 2006-05-17 02:39
.subversion
lrwxrwxrwx 1 bin ourproduct 8 2007-04-30 13:43 trunk ->
1.0.0037
lrwxrwxrwx 1 bin ourproduct 8 2007-04-30 13:43 trunk ->
1.0.0037
It's not causing any problems for us, just through I'd bring it up.
Erik Osterman
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel
_______________________________________________
Gluster-devel mailing list
Gluster-devel@xxxxxxxxxx
http://lists.nongnu.org/mailman/listinfo/gluster-devel
