Prometheus Up, Again

Prometheus was mostly working on the ops machine. I tried to Docker it. Prometheus, itself, came up, but none of the cluster.test names resolved.

Getting github working proved its worth: git push on the ops machine, then git pull on another, and voila! all the files I need are there.

Docker actually proved worthwhile here, too: I just ran the prometheusup script and voila! it was running.

A very simple process – and it proved my supposition: It works fine on stack1.cluster.test because name resolution is coming from elsewhere.

Of course, it is BIND configuration, which has hideous interacting options and g*d-awful syntax. The fixes in named.conf:

//      allow-query     { localhost; 192.168.2.0/24; };
        allow-query     { any; };
        recursion yes;
        allow-recursion { any; };

One would think that adding the Docker name server (172.0.0.11) to the list would do it, but no; that’s too easy. Who knows what networking shenanigans are going on? “any” works, so “any” it is. Recursion shouldn’t be a problem with the name server buried inside a VM network, I hope.

Guess what. It’s bed time again.