Bug #71651 Per mountpoint I/O statistics for NFS filesystems
Submitted: 10 Feb 2014 10:08 Modified: 26 Feb 2014 9:30
Reporter: Daniël van Eeden (OCA) Email Updates:
Status: Verified Impact on me:
Category:MySQL Enterprise Monitor Severity:S4 (Feature request)
Version:3.0.5 OS:Any
Assigned to: CPU Architecture:Any

[10 Feb 2014 10:08] Daniël van Eeden
I've got a host with the MySQL datadir on NFS. But MEM only provides graphs for locally mounted filesystems, not for NFS filesystems.

How to repeat:
Add a NFS filesystem to a monitored hosts and see if there are any disk graphs.

Suggested fix:
Add graphs.
[10 Feb 2014 10:44] Daniël van Eeden
It looks like this message is logged for every NFS mount:

2014-02-10 11:41:26,801  WARN [cme.schedule.Scheduler-12:com.mysql.etools.agent.collection.os.BlockDeviceActivityCollector] could not collect disk {com.mysql.etools.inventory.model.os.BlockDevice : ssh:{*REMOVED*}.nfsserver:/vol/vol01/fs01}, because of No such device or address, skipping (only logged once per device):
[10 Feb 2014 14:00] Daniël van Eeden
The filesystems are shown in the "Filesystem Free Space*" graphs, but there is no info in the "Disk I/O" and "Disk Usage" graphs.

The "Network Traffic Overview" graph is not very helpful either. (it sums up all traffic).

I would like something similar to:
iostat -nm 5
nfsiostat 10 99 /my/nfs/mounted/mysql
[26 Feb 2014 9:30] MySQL Verification Team
Hello Daniel,

Thank you for the feature request!