Forum Discussion
Ken_B_50116
Cirrostratus
CirrostratusNo health monitors on GTM pool or servers, but status still "availble". How?
I have a GTM 11.4.1 HF9 pool with no health monitor assigned. There is also no health monitor assigned to any of the servers ("members") in the pool. How do the pool and servers in the pool all have a green "Available" status if there are no health monitors? These servers have a service port of 123 (NTP). Is the GTM probing that port to get the "Available" status? The servers/members in question are not located on an LTM, so I don't see how iQuery could be at work here as mentioned in SOL5431.
Here's the GTM pool config, although I had to strip a lot of identifying info from it:
[root@--:Active:Standalone] config tmsh list GTM pool NTP-TIER1
gtm pool NTP-TIER1 {
fallback-ipv4 x.x.x.x
fallback-mode fallback-ip
members {
---.org:xxxx {
order 0
}
---.org:xxxx {
order 1
}
---.org:xxxx {
order 2
}
}
}
I think I may have found the answer. In this case, stopping the NTP service on one of the servers resulted in it going unavailable/red in the GTM. This seems to indicate that the GTM is indeed making a connection to port 123 to the server. Additionally, with the NTP service still stopped on the server, I added a Gateway ICMP monitor to the server object in the GTM. While I could ping the server, the object was still red/unviable. This must suggest that as long as the port 123 connection was failing, any other monitors were not even considered.
I spend most of my time in LTM so I am always trying make comparisons between GTM and LTM. In LTM when you add a node (server), LTM automatically assigns an ICMP ping monitor to the object. If that doesn't succeed, then the object is always red. But in GTM, the base/default monitor for the object must be the service port you assign to it. So the GTM is quite different in that regard.
Anyhow, going back to GTM, I am now thinking that all new server objects should be added with the * service port, then use a port-specific health monitor.
Ken_B_50116I think I may have found the answer. In this case, stopping the NTP service on one of the servers resulted in it going unavailable/red in the GTM. This seems to indicate that the GTM is indeed making a connection to port 123 to the server. Additionally, with the NTP service still stopped on the server, I added a Gateway ICMP monitor to the server object in the GTM. While I could ping the server, the object was still red/unviable. This must suggest that as long as the port 123 connection was failing, any other monitors were not even considered.
I spend most of my time in LTM so I am always trying make comparisons between GTM and LTM. In LTM when you add a node (server), LTM automatically assigns an ICMP ping monitor to the object. If that doesn't succeed, then the object is always red. But in GTM, the base/default monitor for the object must be the service port you assign to it. So the GTM is quite different in that regard.
Anyhow, going back to GTM, I am now thinking that all new server objects should be added with the * service port, then use a port-specific health monitor.