Lior Goikhburg
2008-Dec-08 13:14 UTC
[Gluster-users] Problem with extended attributes and AFR
Hello all, I'm having a problem with self-healing feature of GlusterFS. I'm running Debian lenny kernel 2.6.26-1-amd64. When one of the servers, which was unavailable during a file copy, returns back online, AFR doesn't work when files are requested. The missing files are created with size of 0. The check on the extended attributes on the files on both servers return empty tables. getfattr -d /srv/export/files/data/file/test1 <empty> The file systems on servers are mounted with user_xattr option: mount /dev/sda2 on / type ext3 (rw,noatime,errors=remount-ro,user_xattr) The set up is a simple testing environment with 2 servers and one client with the following configuration. on both servers: cat /etc/glusterfs/glusterfs-server.vol volume posix type storage/posix option directory /srv/export/files end-volume volume server type protocol/server option transport-type tcp/server subvolumes posix option auth.ip.posix.allow * end-volume on client machine: cat /etc/glusterfs/glusterfs-client.vol volume client1 type protocol/client option transport-type tcp/client option remote-host 192.168.1.176 option transport-timeout 10 option remote-subvolume posix end-volume volume client2 type protocol/client option transport-type tcp/client option remote-host 192.168.1.177 option transport-timeout 10 option remote-subvolume posix end-volume volume afr1 type cluster/afr subvolumes client1 client2 end-volume I'm mounting the volume on client with: glusterfs -f /etc/glusterfs/glusterfs-client.vol -n afr1 /var/www/storage/ mount glusterfs on /var/www/storage type fuse (rw,nosuid,nodev,allow_other,default_permissions,max_read=1048576) glusterfs -V glusterfs 1.3.12 built on Sep 17 2008 22:09:25 Repository revision: glusterfs--mainline--2.5--patch-797 I'd appreciate it, If anyone can explain me what is causing this problem and how to solve it. Thanks in advance. Lior.
Krishna Srinivas
2008-Dec-08 14:02 UTC
[Gluster-users] Problem with extended attributes and AFR
Lior, Can you see if - getfattr -d -m ".*" <file> dumps anything before selfheal? can you confirm if xattrs are working properly by using setfattr/getfattr commands? If everything looks fine, define "option debug on" in AFR subvolume and start the glusterfs client with "-L DEBUG" and send the logs that get logged when you "cat" the file. Thanks Krishna On Mon, Dec 8, 2008 at 6:44 PM, Lior Goikhburg <glior at hh.ru> wrote:> Hello all, > > I'm having a problem with self-healing feature of GlusterFS. I'm running > Debian lenny kernel 2.6.26-1-amd64. > When one of the servers, which was unavailable during a file copy, > returns back online, AFR doesn't work when files are requested. The > missing files are created with size of 0. The check on the extended > attributes on the files on both servers return empty tables. > getfattr -d /srv/export/files/data/file/test1 > <empty> > > The file systems on servers are mounted with user_xattr option: > mount > /dev/sda2 on / type ext3 (rw,noatime,errors=remount-ro,user_xattr) > > The set up is a simple testing environment with 2 servers and one client > with the following configuration. > > on both servers: > cat /etc/glusterfs/glusterfs-server.vol > > volume posix > type storage/posix > option directory /srv/export/files > end-volume > > volume server > type protocol/server > option transport-type tcp/server > subvolumes posix > option auth.ip.posix.allow * > end-volume > > > on client machine: > cat /etc/glusterfs/glusterfs-client.vol > > volume client1 > type protocol/client > option transport-type tcp/client > option remote-host 192.168.1.176 > option transport-timeout 10 > option remote-subvolume posix > end-volume > > volume client2 > type protocol/client > option transport-type tcp/client > option remote-host 192.168.1.177 > option transport-timeout 10 > option remote-subvolume posix > end-volume > > volume afr1 > type cluster/afr > subvolumes client1 client2 > end-volume > > I'm mounting the volume on client with: glusterfs -f > /etc/glusterfs/glusterfs-client.vol -n afr1 /var/www/storage/ > > mount > glusterfs on /var/www/storage type fuse > (rw,nosuid,nodev,allow_other,default_permissions,max_read=1048576) > > glusterfs -V > glusterfs 1.3.12 built on Sep 17 2008 22:09:25 > Repository revision: glusterfs--mainline--2.5--patch-797 > > I'd appreciate it, If anyone can explain me what is causing this problem > and how to solve it. > Thanks in advance. > > Lior. > > _______________________________________________ > Gluster-users mailing list > Gluster-users at gluster.org > http://zresearch.com/cgi-bin/mailman/listinfo/gluster-users >