Monitor Other User Mount Point

[thefuzz4]

So I have an amazon cloud drive account and I'm using the acd_cli to mount the drive under a user. So naturally only this user can see this mount. Heck even root can't see the mount.

So how would one go about monitor this mount point. I have times where the mount becomes unresponsive and requires a little bit of love in order for it to come back online and I would like to just automate that instead having to do manual intervention. Thank you all in advance for your help with this.

[zalums]

Hi,

I personally don`t have a Amazon Cloud drive so my question is just out of nowhere, but:

Is there any log file which writes state of mounted drive? if yes, then could monitor that for specific lines.

[LenR]

How does the user know the state of the mounted filesystem?

Maybe cron a script that uses sudo as that user to touch a file on the mount? Send the result to Zabbix with zabbix_sender. It might hang if it fails, so I'd add a nodata() check and a dependency on host availability.

[thefuzz4]

Zalimus there is but sometimes the errors that I'm catching already with splunk are actually false.

LenR I think that you have the right idea with that so I'll go investigate how to construct that for zabbix_sender and get it completed. Thanks.

[thefuzz4]

So after some tweaking this is what I came up with in case anyone else wants to use it

#!/bin/bash

######################
#acd_cli check script#
#Written by thefuzz4 #
# 06/19/2016         #
######################

HOSTNAME=yourhostgoeshere
CONFIG=/etc/zabbix/zabbix_agentd.conf


if ! timeout 10s touch ~/acd-sorted/test ; then
#If we're not able to touch the file then lets just quit now and tell zabbix we died
  zabbix_sender -c ${CONFIG} -s ${HOSTNAME} -k mount.acd_cli -o 0 #> /dev/null
  exit 0
fi
#We touched it so lets tell zabbix that we're ok and delete the file
zabbix_sender -c ${CONFIG} -s ${HOSTNAME} -k mount.acd_cli -o 1 #> /dev/null
rm ~/.acd-sorted/test
exit 0

Then on zabbix I have a trigger action to unmount the mount, delete the nodes.db file, do a fresh sync and then mount the file system back up. I might end up putting the commands into a script and just let zabbix call the script instead, but for now its working like it should.