Chris @ Veerotech
2014-Sep-03 23:56 UTC
[Gluster-users] Slow Read/Write Speeds | Fuse Error
Looking for some guidance, I have a cluster set up with three servers for testing. I have two Gluster volumes running, one is distributed replicate (gv01) the other is just distributed (gv02). For gv01, I have two disks on each server setup as bricks. The other volume, gv02 has a single disk on each server as a brick. I am running oVirt and have tied these volumes into it. I have three VM's created and running through oVirt, two using the gv02 and the other using gv01. If I run the VM on the same server the first brick of the volume is set up, the speeds are +100mb/s. If I then move the VM to another server, the speeds drop down to under 10mb/s. The command I am using to test is as follows: dd if=/dev/zero of=test bs=1M count=500 conv=fdatasync I have set up the volume to report out "debug" mode, and this keeps popping up, does it have something to do with the slow speeds: [2014-09-03 23:52:02.301568] D [fuse-bridge.c:48:fuse_invalidate] 0-fuse: Invalidate inode id 140382182774028. [2014-09-03 23:52:17.776083] D [client-handshake.c:185:client_start_ping] 0-gv02-client-0: returning as transport is already disconnected OR there are no frames (0 || 0) Gluster volume options are as follows: [root at u6 ~]# gluster volume info Volume Name: gv01 Type: Distributed-Replicate Volume ID: 0fa1a3e9-12b1-4a21-9310-4e27bed55b54 Status: Started Number of Bricks: 3 x 2 = 6 Transport-type: tcp Bricks: Brick1: u5:/export/sdb1 Brick2: u6:/export/sdb1 Brick3: u9:/export/sdb1 Brick4: u5:/export/sdc1 Brick5: u6:/export/sdc1 Brick6: u9:/export/sdc1 Options Reconfigured: performance.write-behind-window-size: 512MB performance.read-ahead: off diagnostics.client-log-level: DEBUG diagnostics.latency-measurement: on performance.cache-size: 1GB performance.io-thread-count: 64 network.remote-dio: enable user.cifs: disable auth.allow: 162.x.x.* storage.owner-gid: 36 storage.owner-uid: 36 Volume Name: gv02 Type: Distribute Volume ID: 33dea48e-4754-42a9-aeee-96a99c407924 Status: Started Number of Bricks: 3 Transport-type: tcp Bricks: Brick1: u5:/export/sdd1 Brick2: u6:/export/sdd1 Brick3: u9:/export/sdd1 Options Reconfigured: cluster.choose-local: true diagnostics.client-log-level: DEBUG storage.owner-gid: 36 storage.owner-uid: 36 auth.allow: 162.x.x.* user.cifs: enable nfs.disable: off Why are the speeds so slow? Chris -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://supercolony.gluster.org/pipermail/gluster-users/attachments/20140903/3fcdaffb/attachment.html>