I'm performing some tests with GLB services for 2 sites each one containing one vADC box running Pulse VTM 19.3 and one appliance host attending an HTTP service on TCP/80 and TCP/443. For each vADC I setup the pools with one back-end node containing the appliance host, virtual servers for HTTP 80, 443, other 2 VSs for DNS TCP and UDP containing the build-in DNS, a TIG for DNS VSs and another TIG for HTTP VS.
When the back-end node that serves the HTTP services goes down I get the error messages:
"<traffic manager IP address> can't contact its backend nodes and will not raise any IP addresses in this Traffic IP group" on Home tab
"ERROR: Traffic IP address <IP Address> is not hosted on any traffic managers, or traffic managers could not be contacted" for both TIG on Services -> traffic IP group tab.
Any thoughts on how to avoid this behavior when all the back-end nodes go down?
> Any thoughts on how to avoid this behavior when all the back-end nodes go down?
If that's what you need to do, you can create a dummy pool, either a remote DC as a second local pool, or adding a local pool with 127.0.0.1:9090. However, I'm not sure that's going to help directly.
If this is custom global load-balancing, there are some other resources that might be useful:
The desired behavior: when all the back-end nodes go down in the same vADC site the GLB services should fail the same location and redirect the requests to the remote vADC/site.
I added a local pool with 127.0.0.1:9090 on each vADC as per your suggestion. I performed a couple of tests and it's working.