increase respond time when a node down

[thangln]

Hi all,

I'm a newbie and I have some problems, please help me! Thanks in advance!

I want to increase respond time when a node goes down. For example as soon as it goes down, the monitor event will notify me

I've changed the ICMP timeout downto 200ms but the fastes time when monitor event detect a node goes down is about 1-2 minutes 😞

Please help me!

[ceige]

Changing the ICMP timeout will not do this.  What you want to do is change the Polling interval for icmp.  This can be done on a device by device basis or globally through System manager.  To do it for an individual node you would open system manager and right click on the node selecting node details.  To do it globally you will want to open system manager Select File->Network Performance Monitor Settings and select the Polling tab.  From there change the Default Poll Interval for Nodes and apply those settings to all.  One thing to keep in mind is that as you poll more frequently, the max number of elements you can poll on a single polling engine will decrease.

[thangln]

Hi ceiqe,

Thanks for your reply. I've done all things you said before, but the problem is still exist, I means I do before your post :D,  so do you have any other solutions? Thank you!

[Yann · Accepted solution]

Hi,

You might want to increase the Node Warning Interval to something higher than 2 minutes:

In System Manager:

File > Advanced Settings...

Select the "Node Warning Interval".

Change the settings and click on Ok.

HTH,

Yann

[thangln]

Hi all,

Thank for all of your supporting. I've done all things you said, and I think I've reached the limit of NPM. May be it can't respond faster ^_^

Btw, thank you very much!

[ceige]

Ahhh if your looking for the most rapid notification of a node down here is what you do.  In service manager settings change the icmp polling interval to 10 seconds. apply that to all.  Be aware that the faster you poll the nodes the fewer you will be able to support on a single polling engine.  After you set your nodes to 10 second icmp polling, change your warning interval to 0.  This will mark a node down on the first missed ping... beware.  At this point your devices should be marked down after 1 missed ping and sending that ping every 10 seconds.  Now in your alert change the frequency of the check to 15 or 30 seconds and you should be notified in under a minute of any down node. (depending on your choice of notification methods)

Note: In this case any packet loss would be considered a node down and you would get alot of up/down notifications.