RMA connectivity test

Stoltze · Post by **Stoltze** » Thu Nov 19, 2009 11:48 pm

Hi,

A wish from me..

When I need to test, that an active RMA client is connected, I do a "ping localhost". But, besides from being one more test, test is still status Good, even though one RMA in a cluster is down

Using View->Active RMA status, it's possible to see which active RMA's are online

What about an actual test to see this?
Test by = HostMonitor
Test method = Active RMA connected
Drop down to choose a configured active, perhaps even the possibility to choose a cluster
Status is bad after xx seconds of no connectivity
(Or define how often AHM should check the status, and this way Alert profiles works as usual.. Or use number of seconds down = recurrences)

I mean, AHM allready know the status of active RMA's, basically no need to add an extra check. And this test would be much faster to react..

Scenarios:
1 single active RMA = 1 test
1 test that can be used as a master test for everything being checked by this RMA

2 active RMA in a cluster = 3 test
1 test that is okay, when just one of the RMA's is connected. Mastertest for everything being checked by this RMA cluster
1 test that is okay, when RMA 1 is connected
1 test that is okay, when RMA 2 is connected

This way, I allways knows the status of the RMA's.. And I could create a view, with "criteria = RMA test" and there see the status of all active RMA's

Today, I need to do this, to get somewhat same result for a cluster:
1 test that pings localhost. OK if just one RMA is connected
1 test that pings RMA 1 host
1 test that pings RMA 2 host
But there can be a lot of reason for ping to fail, even though the RMA is connected

Does this make any sense..?

KS-Soft · Post by **KS-Soft** » Fri Nov 20, 2009 12:04 am

When I need to test, that an active RMA client is connected, I do a "ping localhost". But, besides from being one more test, test is still status Good, even though one RMA in a cluster is down

I don't think so. If HostMonitor cannot connect to agent, it will set "Unknown" status, not "Ok" or "Host is alive" status (exception - you are using Backup RMA for this agent).
However if you want to use this test as master test for other tests performed by ACTIVE RMA, this may lead to some delay - HostMonitor will not perform test if agent was connected but lost connection recently. HostMonitor may wait several minutes for reconnection.
That's why implemented new trick, there is description
http://www.ks-soft.net/cgi-bin/phpBB/vi ... hp?p=24692

Today, I need to do this, to get somewhat same result for a cluster:
1 test that pings localhost. OK if just one RMA is connected
1 test that pings RMA 1 host
1 test that pings RMA 2 host
But there can be a lot of reason for ping to fail, even though the RMA is connected

A lot of reason for "ping localhost" to fail? I do not see such reasons

unless winsock DLLs are totally destroyed. Then system is unavailable and you should start alert anyway.

Regards
Alex

Stoltze · Post by **Stoltze** » Fri Nov 20, 2009 12:21 am

KS-Soft wrote:
When I need to test, that an active RMA client is connected, I do a "ping localhost". But, besides from being one more test, test is still status Good, even though one RMA in a cluster is down
I don't think so. If HostMonitor cannot connect to agent, it will set "Unknown" status, not "Ok" or "Host is alive" status (exception - you are using Backup RMA for this agent).

When I wrote cluster, I did mean backup RMA..

KS-Soft wrote:However if you want to use this test as master test for other tests performed by ACTIVE RMA, this may lead to some delay - HostMonitor will not perform test if agent was connected but lost connection recently. HostMonitor may wait several minutes for reconnection.
That's why implemented new trick, there is description
http://www.ks-soft.net/cgi-bin/phpBB/vi ... hp?p=24692

Did not see this thread before now. Seems like that will do the trick..!

So, wouldn't this be the correct way to do it then, with version 8.28:
1 test that pings localhost. OK if just one RMA is connected
1 test that use RMA1 and ping "rma itself"
1 test that use RMA2 and ping "rma itself"

This way, I should be able to detect single RMA's not connected, and use a "ping localhost" as mastertest when using backup rma's

Correct?

KS-Soft · Post by **KS-Soft** » Fri Nov 20, 2009 2:21 pm

1 test that pings localhost. OK if just one RMA is connected - TestA
1 test that use RMA1 and ping "rma itself" - TestB
1 test that use RMA2 and ping "rma itself" - TestC

Correct.
Just 1 note regarding TestA: such master test may work with delay. HostMonitor will not perform test if Active agent was connected but lost connection recently (well both of them in case of primary-backup RMA configuration). HostMonitor may wait several minutes for reconnection.
The same is correct for other (dependant) test items performed by the same agent. So its not a big problem.

However in some cases (e.g. you want to start alerts immediately if both agents disconnected) you may prefer to use 3rd auxiliary test item (TestA) configured as dependant on TestB and TestC or use "advanced mode" action assigned to TestB and TestC instead of TestA.

Regards
Alex