Forum Discussion
Nimbostratusgeneric hosts in gtm pools
If generic hosts are added to a gtm pool, how does GTM get health of the node? Is it through polling from one of the self ips? If an alias is added to a pool instead of nodes, how does GTM get the health of the added member?
11 Replies
Cory_50405Noctilucent
GTM will monitor the health of pool members based on the monitor you have assigned to the pool in question. Or if no health check is assigned at the pool level, it'll use the monitor applied to the server objects themselves. Monitor traffic is sourced from a non-floating self IP.
Alias configurations on a wide IP have nothing to do with server objects and monitoring. If you attempt to add a pool member by hostname/FQDN, the GTM will resolve that to an IP address and add it as a member.
jaked_150553I added a pool with an alias in it, there are no virtuals in the pool. The pool comes up green but nothing shows up in the tcpdump for the alias' ip, does it mean the configuration is improper? "Monitor traffic is sourced from a non-floating self IP" - what if the pool member is not among any of the subnets the self-ips belong to? Do we need addition of route to the pool member?
- Cory_50405
If the pool member is not in the same subnet as the self IP, then GTM will use its routing table to determine where to send the traffic to. Do you have a default route specified on the GTM? If so, it'll use that. If not, you may need to configure one.
GTM will select a non-floating self IP as its source.
Ensure you configure your GTM pool with virtual servers so GTM has somewhere to resolve wide IP requests to.
- jaked_150553
I added a generic host and and route to it as there are no available corresponding self ip subnets. But the server is being marked up and down, i see something like this in the logs:
Monitor instance /Common/https xx.xx.xx.xx:443 UP --> DOWN from yy.yy.yy.yy (no route) I also see in the logs that this generic host is being checked by self-ips of other bigips added as bigip redundant sytems and being marked down by them. Any idea how to fix this?
- Cory_50405
This is normal behavior if you have your GTMs setup in a sync group. You can control which GTM devices perform monitoring on the generic host by using prober pools.
http://support.f5.com/kb/en-us/products/big-ip_gtm/manuals/product/gtm-implementations-11-1-0/18.html
Or just setup the proper routing on your other GTM appliances and that should fix your issue as well.
- jaked_150553How can this be achieved in an environment where there are no LTMs, only bigip GTM devices?
- Cory_50405If I understand correctly, your issue at the moment is related to at least one of your GTMs not being able to route to the generic host in question. I would suggest putting a default route in place on each of your GTMs in your sync group. LTM isn't necessary to do the routing or prober pools.
- jaked_150553I tried adding a static route on each of the GTM devices and still see the same issue. I see the (no route) message still coming up in the logs. I also see that other LTM systems already added to the GTM also keep polling the generic host server. First comes the 'no route' message and then comes the 'no enabled virtual server' message, goes red. them after sometime i see 'success' from one of the other LTM systems and then green. State toggles between green and red.
- jaked_150553
Thanks a lot Cory for your help. I figured it out, I had missed adding GTMs in the prober pools. Adding them fixed the issue.
- Cory_50405Good to hear.
