[sheepdog] 3 nodes cluster - always something to fix when checking

Petre Bandac petre at kgb.ro
Sun Jan 18 13:41:04 CET 2015


Hello

I am having the following problem: after creating a 3 node cluster and
importing a few images, every time I issue a check command (per cluster
oder per vdi), it has something to fix and repair.

I suppose this is not the normal behavior, but since the vm's are
up'n'running, I see no explanation; before converting these
medium-sized images, I have played around with 2 GB images, and there
no fix was needed and the check was instant (without the status line
slowly growing up to 100%).

My question is: when all the nodes of the cluster are running, wouldn't
it be normal at checking to be nothing found to fix ? and the duration
of the check is direct proportional with the size of the images ?

Thanks,

petre

###################


#dog cluster format -c 2:1

# dog cluster info
Cluster status: running, auto-recovery enabled

Cluster created at Sun Jan 18 02:46:17 2015

Epoch Time           Version
2015-01-18 02:46:17      1 [10.0.0.70:7000, 10.0.0.87:7000,
10.0.0.96:7000]


# dog node list
  Id   Host:Port         V-Nodes       Zone
   0   10.0.0.70:7000      	286 1174405130
   1   10.0.0.87:7000      	66 1459617802
   2   10.0.0.96:7000      	32 1610612746

# dog vdi list -r
= webserver-sys 0 15728640000 15451815936 0 1421543762 e16ca 2:1 
= provisioning 0 23622320128 19889389568 0 1421541995 27eb28 2:1 
= webserver-data 0 53477376000 35131490304 0 1421543833 677471 2:1 
= alex 0 53687091200 12205424640 0 1421543559 cdad97 2:1 

sheepdog version -->  0.8.3-1~bpo70+1 on debian 7 nodes

every check command (below with the machines stopped, after the
conversion) gives the following output:

dog cluster check
fix vdi webserver-sys
100.0 %
[================================================================================================================================]
15 GB / 15 GB finish check&repair webserver-sys fix vdi provisioning
 22.7 %
[=============================>
] 5.0 GB / 22 GB     object 27eb28000002dd is inconsistent 100.0 %
[================================================================================================================================]
22 GB / 22 GB finish check&repair provisioning fix vdi webserver-data
100.0 %
[================================================================================================================================]
50 GB / 50 GB finish check&repair webserver-data fix vdi alex
100.0 %
[================================================================================================================================]
50 GB / 50 GB finish check&repair alex


after this, issuing the same command again gets the same result -
something to repair

# dog cluster check
fix vdi webserver-sys
100.0 %
[================================================================================================================================]
15 GB / 15 GB finish check&repair webserver-sys fix vdi provisioning
 22.7 %
[=============================>
] 5.0 GB / 22 GB     object 27eb28000002dd is inconsistent 100.0 %
[================================================================================================================================]
22 GB / 22 GB finish check&repair provisioning fix vdi webserver-data
100.0 %
[================================================================================================================================]
50 GB / 50 GB finish check&repair webserver-data fix vdi alex
100.0 %
[================================================================================================================================]
50 GB / 50 GB finish check&repair alex




More information about the sheepdog mailing list