Wrong disk graphs after moving MS cluster node

[Thomas]

Hi CACTI-gurus,

after moving a cluster node from one machine to the other (Windows 2000 server) I have to manually "Reload Data Query" to get the correct disk graphs. The Re-Index Method has been set to "Verify all fields".
Does/will CACTI have a method to support cluster node moves?

Thanks in advance

[robn]

I have the same problem, has anyone got any surgestion/fixes.

It possible to fix it (I think) it, but will slow thing a little
Check the label of the device - if it changed find out and update the location of the HD. If I get time I might have a gander at the code and see if I can work out how it works and what needs to be changed to get it to cope with cluster fail overs and the adjusting of HDs.

[BSOD2600]

If you click the green circle next to the SNMP mounted partions in the associated data query section, cacti should reindex it for you.

[robn]

Yeh - I discovered that.

Just could do with the system to do it automatically, espically due to the wide range of HD sizes in the cluster - from 20gb all the way upto 250gb. You end up with some pretty messed up graphs at times.

[zuessi]

Hi there,

I did experience the same issue. Now I am collecting the information from the cluster-resource instead of the individual host. This provides accurate for the disk-space. (Other parameters like CPU or network traffic is still taken from the individual hosts.)

Have fun

[robn]

Bit of a delay in the reply, but its been a busy start to the year.

Same thing happens for the cluster resource - windows doesn't pad out the disk index in snmp, so when you fail over a resource with a disk the disk index gets redone, and hence I need to re run the verbose query on "SNMP - Get Mounted Partitions".

I suppose if a script/no autenticated page could be made I could get windows to run it when said cluster resource fails.

Either that or try to make a new data source based on ngclient data - only problem is that data isn't the same (ones it bits the others in megabytes if I rember right).

[BSOD2600]

If this is happening a lot, maybe you should look into running the poller_reindex_hosts.php script once a night.

[fwiethof]

hi!

@zuessi or other ppl: we tried that with one of our customer's clusters, and i was not sucessful in polling the cluster since the snmp packets are being answered from the node's ip address instead of the resource's ip address.

any hints on the server configuration to handle that?

[robn]

Running poller_reindex_hosts.php does the trick. I've made various php pages for each of the clusters which runs this script for each virtual server. Just need to find some time to writing a script to get windows to fetch this page on fail over.