Tweaking Polling Tasks

[Anthony_67]

Im trying to make my CIM7 SP2 system more responsive to system events. What polling tasks can I tweak so I know quicker if a system component has failed?

What polling task does the hardware checks? (ie hard disk failure, CPU hot etc)

I changed the SNMP status polling for servers from the 5 minute default to 2 minutes - will this cause any issues apart from increased traffic??

Thanks
Anthony

[Jeff Mathews]

Anthony,

You can build a custom notification task that will notify you on any of these right away. What you need to do is (lets say for the disk failure notification):

1. On the task pane, select "new notification task" and "email notification".

2. Type your to address, and a subject line for the notification email, and press next.

3. Give you task a name, like "Disk status notification".

4. Select the "new query" button.

5. Select the "events by severity" radio button.

6. Select the "Event(s) of type" check box.

7. Select the "Type" hyperlink in the bottom pane.

8. Open the "Storage" folder, and scroll down and select all the "Disk ...." events, and the "Physical Drive Status change" check boxes. Press Ok.

9. If you want, you can now select the "Device(s) of type" check box, then click the "Type" link beside devices in the bottom pane.

10. Select the "servers" check box and press ok.

11. Give you custom query a name like "Server drive status" and click save.

12. Now click the schedule link in the bottom pane, and click the "when new devices or events meet your query criteria" radio button at the top. Press ok.

13. Save you notification task and you should be set.

This type of notification should be sent as soon as the SNMP trap is received from the server in question.

Jeff

[David Claypool]

From an SNMP polling perspective, the Insight agents have a single SNMP OID called "MIB Status Array" which is the only thing that is queried during the SNMP Status Polling task. This is different from many SNMP systems that must query each individual item to get status. That means that IM7 is very conservative in the overhead imposed on the network by doing frequent status polling.

The MIB Status Array contains overall status for all health items. You;ll have to drill into the agent to get detail. Polling is a general health check. An SNMP trap originated by the agent when it detects a problem will contain detailed information.

[Rob Buxton]

I reduced my SNMP Status Polling For Servers down to 1 minute.

I was finding that some of the new Proliants can reboot so fast they were not being caught as unavailable.

The IM Server currently has several other applications, it is running the Proliant PMP and has not shown any performance problems.

[Anthony_67]

Hi Rob

I set my setting also to 1 minute. Do I also need to reduce the data collection interval on the Servers (Under SNMP in the HP Management Agents)??

Im assuming these Agents are what detect the issue in the first place - would that be correct?? Would setting this to 1 minute cause much overhead.

Thanks
Anthony

[Rob Buxton]

Must confess I should've read the first post a bit more closely.

To be more responsive to Agent Generated events then yes, I'd say you need to change the interval on the Agent Servers. I have not done this so cannot comment on performance degradation.
I can't see that it would impact the IM Server as, in theory, that won't be seeing any more traffic.

As I said, I changed the polling interval on the IM Server, and that was to be more responsive to detecting outages.