How monitoring many mounted partitions

[stefano]

Hello,

maybe this can seems a simple question, but it isn't. I have about 100 servers that have many different partitons. My needed it to monitor every partition without loading manually each one and alto so get warning for every has free space between 10 and 5% and alert for every with less than 5% free space.

It's possibile in zabbix?

Thanks
Ste

[unizabbix]

I have the same problem, someone has found the way to do this?

Bye

[xs-]

short answer: no

long answer: all unique partitions or are there multiple filesystem layouts which are shared by different hosts? You could try and group hosts which have a similar layout and create a couple of filesystem templates.

[trikke]

Hi Ste,

if your monitoring Unix server u could create a item like:
system.run[df -lkh|sort -rk 5|awk '{print $5":"$6}'|xargs]

this gives u a list with the Filesystem using the most space on top.

Using this trigger:
{Template_Solaris:system.run[df -lkh|sort -rk 5|awk '{print $5":"$6}'|xargs].iregexp((9[0-9])%)}=1

sets to True if there is a string 90% to 99 % found.

Thats the way i do it on Solaris, I guess something similar in Windows is possible using a windows command, which makes a list of your partitions!

Greets
Patrick

[unizabbix]

In this way you are alerted only when one of the filesystem goes up the limit (90 or 99).
For example :
if i have 5 filesystem mounted on :
1. /mnt/part1
2. /mnt/part2
3. /mnt/part3
4. /mnt/part4
5. /mnt/part5
and 1 is > 90% and 2 > 99% the trigger goes true. If, in the meantime, 3 becomes greater 90% the triggers doesn't alert me becuse is already true.

How can I receive alert everytime a filesystem goes up the limit?

[trikke]

at the trigger definition u select "Normal + Multiple TRUE Events".
This will send an event every item-intervall untill the situation is cleared for all fileystems!
Greets
Patrick

[unizabbix]

at the trigger definition u select "Normal + Multiple TRUE Events".
This will send an event every item-intervall untill the situation is cleared for all fileystems!
Greets
Patrick

mmm....i have zabbix 1.4.5! Ok, i have to upgrade.

Thanks a lot

[sege]

This won't solve graphing the different systems which I'd like to have as well.

And if the partition is 100% filled the trigger would say false? That's not ok?

Is there really now way of discover partitions like once an hour and automatically add them to the database? If so, how can I do this?

[jjsmith]

Is there really now way of discover partitions like once an hour and automatically add them to the database? If so, how can I do this?

There is... but it's a little complicated.

1. Create a script which uses the "df" command and reports information via snmp. Use this script on each computer you wish to monitor.

2. Create a script which queries each computer via snmp and creates items to be monitored and triggers. Output the information into a Zabbix readable XML file.

3. Use some script to automatically import the information into Zabbix.

Both #2 & #3 can be run on a single machine (i.e. Zabbix server)

I told you it's tricky. I've done #1 & #2 (less triggers), but #2 is a little messy Reply to the topic if you want the code.

[jjsmith]

Oh, I'll also clean up my code if anyone wants it. & I wrote these scripts in perl (+used others' code) for monitoring Linux machines.
Script #1 is the only one that would need to be changed for monitoring windows, etc. Plus, it is possible to monitor multiple types of machines using the same SNMP directives. This would make implementation across diverse architectures more uniform.