Ernie Dunbar
2014-Sep-04 22:18 UTC
[Gluster-users] Gluster appears to be connecting to its peers, but filesystem remains unchanged on the other peer.
Hi, I just started using Gluster today to build a new fileserver, and so far I'm impressed with the ease of set-up and configuration. However, my cluster appears to be working normally, yet no updates are made on the actual filesystem. I change files on one node, and nothing shows up on the other. Since I'm new to this, I don't know what kind of troubleshooting steps to take yet, but I've executed a few commands on the gluster cli: root at nfs1:/nfs# gluster volume profile nfspool info Brick: nfs1:/nfs ---------------- Cumulative Stats: %-latency Avg-latency Min-Latency Max-Latency No. of calls Fop --------- ----------- ----------- ----------- ------------ ---- 7.32 30.00 us 30.00 us 30.00 us 1 GETXATTR 18.54 38.00 us 25.00 us 51.00 us 2 STATFS 74.15 76.00 us 61.00 us 94.00 us 4 LOOKUP Duration: 7663 seconds Data Read: 0 bytes Data Written: 0 bytes Interval 0 Stats: %-latency Avg-latency Min-Latency Max-Latency No. of calls Fop --------- ----------- ----------- ----------- ------------ ---- 7.32 30.00 us 30.00 us 30.00 us 1 GETXATTR 18.54 38.00 us 25.00 us 51.00 us 2 STATFS 74.15 76.00 us 61.00 us 94.00 us 4 LOOKUP Duration: 7663 seconds Data Read: 0 bytes Data Written: 0 bytes Brick: nfs2:/nfs ---------------- Cumulative Stats: %-latency Avg-latency Min-Latency Max-Latency No. of calls Fop --------- ----------- ----------- ----------- ------------ ---- 7.11 31.00 us 31.00 us 31.00 us 1 GETXATTR 20.64 45.00 us 42.00 us 48.00 us 2 STATFS 72.25 78.75 us 64.00 us 91.00 us 4 LOOKUP Duration: 7661 seconds Data Read: 0 bytes Data Written: 0 bytes Interval 0 Stats: %-latency Avg-latency Min-Latency Max-Latency No. of calls Fop --------- ----------- ----------- ----------- ------------ ---- 7.11 31.00 us 31.00 us 31.00 us 1 GETXATTR 20.64 45.00 us 42.00 us 48.00 us 2 STATFS 72.25 78.75 us 64.00 us 91.00 us 4 LOOKUP Duration: 7661 seconds Data Read: 0 bytes Data Written: 0 bytes root at nfs1:/nfs# gluster volume status Status of volume: nfspool Gluster process Port Online Pid ------------------------------------------------------------------------------ Brick nfs1:/nfs 49152 Y 2318 Brick nfs2:/nfs 49152 Y 1833 NFS Server on localhost 2049 Y 2882 Self-heal Daemon on localhost N/A Y 2335 NFS Server on nfs2 2049 Y 1985 Self-heal Daemon on nfs2 N/A Y 1953 There are no active volume tasks gluster> peer probe nfs2 peer probe: success: host nfs2 port 24007 already in peer list gluster> peer status Number of Peers: 1 Hostname: nfs2 Uuid: ab13df7b-d7e7-46c9-8c43-c347b68a2a08 State: Peer in Cluster (Connected) gluster> peer status nfs2 Usage: peer status gluster> peer status Number of Peers: 1 Hostname: nfs2 Uuid: ab13df7b-d7e7-46c9-8c43-c347b68a2a08 State: Peer in Cluster (Connected) gluster> volume list nfspool gluster> volume sync Usage: volume sync <HOSTNAME> [all|<VOLNAME>] gluster> volume sync nfs2 all Sync volume may make data inaccessible while the sync is in progress. Do you want to continue? (y/n) y volume sync: success To me, this looks like everything is working and normal. I can probe each peer from each other peer, and pings work as well. This tells me there shouldn't be any split-brain condition either. Yet: root at nfs1:/nfs# ls file1 file12 file15 file18 file20 file23 file3 file6 file9 file10 file13 file16 file19 file21 file24 file4 file7 lost+found file11 file14 file17 file2 file22 file25 file5 file8 and on the other node: root at nfs2:/nfs# ls foo lost+found -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20140904/eaf1c994/attachment.html>
Kaleb KEITHLEY
2014-Sep-04 23:31 UTC
[Gluster-users] Gluster appears to be connecting to its peers, but filesystem remains unchanged on the other peer.
On 09/04/2014 06:18 PM, Ernie Dunbar wrote:> Hi, I just started using Gluster today to build a new fileserver, and so > far I'm impressed with the ease of set-up and configuration. > > However, my cluster appears to be working normally, yet no updates are > made on the actual filesystem. I change files on one node, and nothing > shows up on the other.You shouldn't be writing anything directly to the backing (brick) file system(s). Mount the gluster volume somewhere. It can even be on one of the servers, e.g. `mount -t glusterfs nfs1:nfspool /mnt`. Then write your files to /mnt/.... The files will then be written on both servers. -- Kaleb