Custom Alerts reacting strangely

[VTorque]

Hi All,

I have had a go at configuring my first custom alerts, with some unexpected results. Basically, I need an alert in the event that a host drive space useage exceeds 80%.

Anyway, I created a definition at the "hosts and clusters" level in VC. I configured a trigger for:

\- "Host disk usage"

with the condition

\- "Is Above"

\- % alert = 80

Under the reporting tab, the tolerance was set to 5%, and the frequency was set to 60 minutes.

Under the actions tab, we set the action to "send an email" - we have an SMTP address for our pagers and this was entered.

The general idea was that a pager alert would be received if any host in the environment had disk usage over 80%. We would not receive an alert for 60 minutes after the previous one.

The behaviour exhibited was quite strange.

As soon as the alert was enabled, we started to receive a page from each host (3 of them) about every 2 minutes. VC reported no triggers (as it would normally for say CPU usage or whatever) ie: under alarms => triggered alarms, there was no entry. However under tasks and events I could see where each host had triggered the alarm, and sent the email.

The hosts have plenty of free space when I check with the df -h command.

Has anyone noticed this also? Or have I configured something incorrectly?

[radhika1780]

The question here is answered in two steps.

1) You are receiving page for every 2 minutes, because VC 2.0.1 has a bug reporting to frequency not working properly though set for any duration.

That means if you have set frequency to 60 minutes, alert ignore this setting, and send page whenever it is triggered.

2) Yes, you have done correct configuration for custom alert. As the values keep changing frequently, sometimes alarms cannot be seen in Triggered alarms section. But since alarm is triggered and if it has appeared in Tasks & events tab, then that would be fine.

Message was edited by:

radhika1780

[VTorque]

Hey - thanks..

But that still does not explain why the alert is triggering. The host is no where near 80% disk useage.

Very strange.

[hicksj · Accepted solution]

The monitor you're setting up is not for disk usage (storage), its monitoring disk utilization (I/O).

View the performance graphs for storage. There is nothing in there to poll for storage capacity. Is your I/O high? That should be why alerts are being triggered now.

If you want storage utilization, you'll need to use something else. Either SNMP monitoring, local scripts on the host, or hardware agents. (we use the HP Insight Agents, and define thresholds in the agent. It then sends an snmp trap upon exceeding the threshold)

[VTorque]

Now that is really interesting!

It was a VMware engineer that suggested I setup the alerts in the manner I described above to alert to disk usage.. hmmm... He has lead me astray before too! anyways...

I think I will go for SNMP monitoring as we have a server for this already..

I will mark your answer as correct as soon as I confirm this tomorrow when I am back in the office. Thanks in advance though!

[VTorque]

Thanks guys!