Add nfsiostat to checkmk for all the workers nodes
In the same way we should track the nfs server with iostat
we need to do the same in all the workers to check the use of NFS we are doing. This will be particularly handy when we move to a distributed filesystem as we will need to know if things are working at the same rate or less.
A sample output of nfsiostat
(consider that the output format is kinda broken for now http://linux-nfs.vger.kernel.narkive.com/tjITNl10/patch-nfsiostat-restore-output-format)
nfsiostat 5 100 /var/opt/gitlab/git-data
10.1.0.21:/var/opt/gitlab/git-data mounted on /var/opt/gitlab/git-data:
op/s rpc bklog
891.58
0.00
read:
ops/s kB/s kB/op retrans avg RTT (ms) avg exe (ms)
24.792
2911.446
117.434
0 (0.0%)
4.410
4.997
write:
ops/s kB/s kB/op retrans avg RTT (ms) avg exe (ms)
1.486
242.703
163.366
0 (0.0%)
10.667
880.263
10.1.0.21:/var/opt/gitlab/git-data mounted on /var/opt/gitlab/git-data:
op/s rpc bklog
659.80
0.00
read:
ops/s kB/s kB/op retrans avg RTT (ms) avg exe (ms)
110.200
26533.862
240.779
0 (0.0%)
4.739
6.820
write:
ops/s kB/s kB/op retrans avg RTT (ms) avg exe (ms)
0.400
1.035
2.588
0 (0.0%)
5.500
6.000
10.1.0.21:/var/opt/gitlab/git-data mounted on /var/opt/gitlab/git-data:
op/s rpc bklog
166.40
0.00
read:
ops/s kB/s kB/op retrans avg RTT (ms) avg exe (ms)
1.200
6.027
5.023
0 (0.0%)
0.500
0.500
write:
ops/s kB/s kB/op retrans avg RTT (ms) avg exe (ms)
0.000
0.000
0.000
0 (0.0%)
0.000
0.000
With this we will know how many iops we are sending from each client to each mount (I'm limiting to one mount in this sample) and we will be able to learn how the system is behaving.