Re: [monit] Monit reporting wrong disk usage on OpenBSD

[Martin Pala]

You should use the mountpoint of the filesystem instead of the device 
... the problem is, that when the block device is used, some operating 
systems report the usage of the filesystem where the "block device" 
special file (e.g. /dev/raidg0) is stored, not the usage of the 
filesystem which is stored on the device (e.g. /home).

So for example use:

--8<--
  check device fshome with path /home
  ...
--8<--


Martin


Jake Conk wrote:
> Hello,
>
> First a little background, I'm using the latest monit at this time
> monit 4.10.1 (compiled, not by package) and I'm running OpenBSD 4.1. I
> cofigured all my disk partitions to be monitored with monit however
> when I looked at the web interface for details I got this for disk
> usage on ALL my disks, "35.1% [29.7 MB]" and for inode usage I got
> "25.6% [3283 objects]".
>
> Here is my service configuration for each device:
>
> # Monitor home(/home) filesystem
> check device fshome with path /dev/raid0g
>         if failed uid root then unmonitor
>         if failed gid operator then unmonitor
>         if space usage > 80 % then alert
>         if inode usage > 80 % then alert
>         group filesystems
> # Monitor root(/) filesystem
> check device fsroot with path /dev/raid0a
>         if failed uid root then unmonitor
>         if failed gid operator then unmonitor
>         if space usage > 80 % then alert
>         if inode usage > 80 % then alert
>         group filesystems
> # Monitor tmp(/tmp) filesystem
> check device fstmp with path /raid0e
>         if failed uid root then unmonitor
>         if failed gid operator then unmonitor
>         if space usage > 80 % then alert
>         if inode usage > 80 % then alert
>         group filesystems
> # Monitor usr(/usr) filesystem
> check device fsusr with path /dev/raid0d
>         if failed uid root then unmonitor
>         if failed gid operator then unmonitor
>         if space usage > 80 % then alert
>         if inode usage > 80 % then alert
>         group filesystems
> # Monitor var(/var) filesystem
> check device fsvar with path /dev/raid0f
>         if failed uid root then unmonitor
>         if failed gid operator then unmonitor
>         if space usage > 80 % then alert
>         if inode usage > 80 % then alert
>         group filesystems
>
> I then thought the problem might be because doing a software raid 1
> with OpenBSD and maybe monit can't properly read a software raid1 disk
> so I created a service check for an actual device on my system:
>
> # Monitor sd0b(/xx) filesystem
> check device fstest with path /dev/sd0b
>         if failed uid root then unmonitor
>         if failed gid operator then unmonitor
>         if space usage > 80 % then alert
>         if inode usage > 80 % then alert
>         group filesystems
>
> Magically that reported the same results as all my raid file systems.
>
> Has anyone else reported the same problems on OpenBSD and is there a fix?
>
> Please let me know.
>
> Thanks,
> - Jake
>
>
> --
> To unsubscribe:
> http://lists.nongnu.org/mailman/listinfo/monit-general