Standby management server disconnected alert.

[Hongyu_Chen]

We have two management servers, mgmt-1 and mgmt-2, both running on Gaia R81.10. mgmt-1 is active and mgmt-2 is standby. mgmt-1 has been configured to send out email alert when it think mgmt-2 is disconnected.

Recently we are receiving much more mgmt-2 disconnected alerts than before, but when we login to mgmt-2 it seems to be ok.

I don’t know how mgmt-1 detect the status of mgmt-2, maybe it is using ping or some other methods. We are investigating the root cause of the alerts.

My question is, while we are investigating,  as a workaround is there a way to increase the probe interval from mgmt-1 to mgmt-2 to reduce the number of alerts as this is very annoying? But I don't want to just simply disable the function because mgmt-1 is also managing the security gateways. I just want mgmt-1 to send fewer mgmt-2 disconnected alerts.

Thank you!

[Chris_Atkinson]

Could you please share the JHF (jumbo) level of the machines, is it recent?

How's the latency between the machines and is there any packet loss?

What is the spec/load of the respective  management machines?

[Hongyu_Chen]

Hi Chris,

JHF is R810.10 Take 66.

We did see sometimes cpu is high.  please see below screenshot

But no packet loss. 

Our priority is to adjust the probing interval if that is adjustable, because the alert email is pouring into the mail box. 

[Chris_Atkinson]

The probing interval you'd need to check with TAC as I'm unsure. You can deffinately set the sync timeout otherwise per sk176165.

[Hongyu_Chen]

Thanks Chris. sk176165 seems apply to MDS while ours is not. 

[_Val_]

@Hongyu_Chen It does not matter, it is also applicable to SMS

[the_rock]

Just as a test, though I dont have mgmt HA, I ran it on single SMS and worked fine, no issues. I was able to also push policy afterwards, so I would say its most likely safe to do in your case as well. I sadly dont have access to my R81.10 lab mgmt, as its being used for something else atm, but I did it on R81.20, though Im positive it makes no difference.

[the_rock]

All valid points by Chris. Also, any relevant logs you can provide about this?

[Hongyu_Chen]

What logs you wanted to see?

[the_rock]

Never mind, my bad, I used the wrong term. You said there are some alerts, but I assumed there might be some logs related to the issue.