Cheshire Cat Computing

Hi

We have 2 ESX-Server, one 2.5.3 and a new one version 3. I am using check_esx2.2.
All the checks except Memory Usage on ESX 3 are ok. Among the Nagios status information I have "CRIT: Balloon drivers in action! (51208.593%)".
This cannot be right, in the ESX console the Ballon drivers is not used, on this just installed server only 1 V-Host is running, but no used at the moment.

Now I'v seen a new version 2.3 is available. I tried this version, but I had 2 problems. The Memory Usage shows no more the details about private, shared, balloon and swap memory usage, only the free and total available memory. Why? Has it been removed?
The second problem, on ESX 2.5.3 the memory usage of the only V-Host had the Nagios status "UNKNOWN" and the status information "Please wait, data being gathered."

So I moved back to version 2.2 with the only problem of the critical Balloon drivers.

Thanks for any help!

Peter

We don't have a v3 ESX server here to test on, so I'm running a bit blind. I know it has different formts to the /proc/vmware tree files (see other thread - this causes the huge balloon problem), so I've been trying to generalise vmware-stats for this new version.

The new version should work on both, although the memory stats wont be reported if they cant be deduced.

Would you volunteer to test a modified version with debug code on your v 3 server?

_________________
Steve Shipway
UNIX Systems, ITSS, University of Auckland, NZ
Woe unto them that rise up early in the morning... -- Isaiah 5:11

I can try it, I hope to have some time. At the moment it sould no be problem, the Nagios monitoring for the ESX Servers are still in my test phase.

I've just heard from another ESX 3.x user and it appears that check_esx2 version 2.3 is working with his system (remember to install the new version of vmware-stats and the snmpd.conf on the ESX server).

The 'please wait, data being gathered' message appears for the first 2 minutes of monitoring as it needs to build up a history before it can calculate rates and so on (particularly for CPU stats).

If the please wait message remains for more than 5 mins then you may have the old 2.1 version of vmware-stats, or else you have some other problem. Try running vmware-stats from the command line on the ESX server and see if it gives you any error messages.

_________________
Steve Shipway
UNIX Systems, ITSS, University of Auckland, NZ
Woe unto them that rise up early in the morning... -- Isaiah 5:11

I've downloaded again version 2.3 and copied the necessary files to the Nagios Server and both ESX Servers. Aferwards I edited the snmpd.conf except the SNMP traps settings, I think I don't need this lines, finally I restarted the SNMP service(s).

But still after hours the "Used Memory" Nagios status of the Virtual Hosts on the ESX 2.x and 3 is "UNKNOWN" with the informaiton "Please wait....". The same result on the shell.

What I am wondering why I don't have anymore the detailed information (pvt/shr/ball/swp) of the "Used Memory" of the hole ESX serves, only free Memory and Total Memory. There is no difference between the old and new ESX Server. Is it in version 2.3 excluded?
Other Nagios checks like "CPU Usage" are ok on both ESX.

The command line output on ESX 3 with vmware-stats looks ok, no errors.

_________________
Steve Shipway
UNIX Systems, ITSS, University of Auckland, NZ
Woe unto them that rise up early in the morning... -- Isaiah 5:11

Ths execute permissions for vmware-stats on both ESX Severs are ok, it gives also some output if I execute it on the shell.
The SNMP daemons are working as well as the vmware-snmpd daemon, but this daemon is only available on ESX 2.5x, in version 3 it seems that it's no more present. I restarted the daemons several times, but without success for the Memery Usage check.

I have some output with snmpwalk, but I don't know if the vmware-stats output is under it. Maybe you can tell me on what I should pay attention.

But again, the Memory Usage with esx2.2 except the "CRIT: Balloon drivers in action!" message on Nagios was ok, therefore I think the SNMP configurations must be alright.
Other checks over SNMP are working, e.g. the CPU Usage. Of course I can remove the Memory Usage check, but for the condition of a ESX Server the detailed memory usage is one of the most important information.

_________________
Steve Shipway
UNIX Systems, ITSS, University of Auckland, NZ
Woe unto them that rise up early in the morning... -- Isaiah 5:11