ESX Snmp Template & Datastore monitoring

[plop28]

Hi,

I have a problem with zabbix 2.2.2 and SNMP Template. It works pretty well with one ESX5.5, SNMP gives me some data (Disks/RAM/CPU/Interface).

For all other ESXi i have (1 ESX5 / 2 ESX5.1 / 3 ESX 5.5) SNMP can't find datastore.

I don't kwow what is wrong with that and why it works with one.

Where should i start to solve my problem ?


Thanks,

[jamesNJ]

On your linux zabbix host make sure you have the proper vmware MIBs installed, and use snmpwalk to dump all the available variables from your server to a file. I think the data stores should be there, but they might be listed by their internal numeric identifier, and not the friendly datastore name you see in vsphere gui.

[plop28]

Hi

I added HOST-RESSOURCES-MIBS but no changes. And i just see that snmp queries doesn't works well. I have some points on my graph instead of a line.
But why datastore are well monitored on one ESXi without theses MIBS..

Now, im using Virt-VmWare. It works well but not as much data as SNMP templates. BTW i had two ESX with same UUID, so i have changed one.

Thanks for your reply

[jamesNJ]

If snmp queries are not working at all, then your ESX server(s) may not have the snmp service running.

Check the ESX documentation. I I am not sure if snmp is disabled by default, however sometimes it is not straightforward to get it working. I think the vsphere GUI has some method of starting/stopping the snmp service, but the docs point to using the command line to configure the service, community string, etc.

[plop28]

Hi James,

SNMP is enabled and SNMP Communities is set as public on all ESX.

Here a Sreenshot that illustrate the kind of data SNMP give me

http://hpics.li/d795f45

[jamesNJ]

I can't tell for sure, but the graph suggests that zabbix is not getting data, implying a possible connection issue.

SNMP v1 and v2 are UDP based, so it is conceivable that zabbix periodic requests are not seeing replies. Or at least that is what I would first suspect.

If you know how to use snmpget and snmpwalk on linux, I might suggest writing a small script that periodically polls for some SNMP value and see if you can observe reliable or intermittent replies. If it is intermittent, you can start by looking at iptables firewall, ESXi firewall, and the network/firewall in between possibly dropping UDP packets.

If you get reliable response then you would probably need to look more closely at what is happening with zabbix.

[plop28]

snmpwalk give me a timeout one out of two or three.

more informations with logfiles, it appears to be related with storage checks :

when i have a timeout with snmpwalk i get a new line:

in syslog.log :
http://pastebin.com/0CdeVEkX

in vmkernel.log
http://pastebin.com/SyWZSLSC

It's apparently not in Zabbix side..

[plop28]

Ok, i try increase SNMP timeout in zabbix_server.conf and it works so far.

again, thanks for your help JamesNJ

[jamesNJ]

I'm glad it was of some help.

However I would suggest following up on those storage messages, it could be an indicator of something more serious. I seem to remember that ESX allowed for some switching of modes and may have an option to disable smart or other checks if you are not interested in them.

[plop28]

Well, this ESX don't have raid card. It runs with Intel Raid and apparently it's not well supported by ESXi OS.