CAn someone help me interpret this Operations Overview Screenshot

[Anon Anon]

[Anon Anon]

Ratings for servers

Servers:

Most are Unavailable, 

172.30.102.100 is normal for 5 minutes then goes to unavailable

What does that mean, was it reachable for only those 5 minutes?

172.31.102.156 goes from unavailable top no data. What is happening here?

Also since there are so many unavailable does that mean that these servers are just not being seen by the monitors (these are GIgaStor  feeds)

What conclusion can I draw from this, that the probes (be it spans or taps) collecting the data are not consistent or are servers bouncing?

 

[Stuart Weenig]

Server availability in ADA is derived. By this i mean that as long as one
SYN/ACK packet or one response packet is seen coming from the server for
one of the applications on the server, ADA will mark that 5 minute time
frame as available. If nothing is seen during the 5 minutes to indicate
that the server is responsive, the collector will do a TCP connect to the
server on the application port(s) monitored. If it is successful, the 5
minute period is marked as 'white'/no data. If the TCP connect fails, the
server is marked as unavailable.

I would start by looking at the network between the collector and the
server to make sure the collector can successfully open a TCP session on
the server without being blocked. If it can't, that's why they're showing
up unavailable.

[Anon Anon]

Thanks for your response Stuart

You helped me form a premise.

I do believe this "unavailable state" of the server is by design.
Upon closer inspection all the server IPs are from server VLANs, so it would make sense to block/discard connections to the servers from non-server subnets (ADA monitors are not in server VLAN!).
Hence a bunch of red, that gets acknowledged.

But periodically communication between servers will happen and I'll get measurements.

This may the the effect of an ACL for security.

Thank you!

 

 

 

 

[Stuart Weenig]

Correct. You have two options: 1) enable connectivity between the
monitoring VLAN and the server VLAN or 2) disable availability monitoring
on servers that will not accept connections from the monitoring VLAN.