TLDR Chris was having troubles with setting an IP for node cluster communication 'openobserve' on-premise. Prabhat recommended reporting the issue on GitHub.
it happens for both http_addr and grpc_addr
How are you setting it up. an etcd cluster. and a binary on each machine
?
Yes, external etcd cluster. For oo currently a series of VMs (1 router, 2 ingesters, 4 queriers, 1 compacter/alerter). Will scale up if it goes full production
There is no direct way to do it now. Give us some time to figure out a solution.
Ok thanks. Optionally Specifying a bind address which is then advertised would be one answer, but of course there are others
Can you create a github issue for this
Ok, finished here now but will do it tomorrow
Chris
Mon, 17 Jul 2023 17:01:24 UTCopenobserve on-prem - just firing up an oo cluster for the first time.... i can see each node logging the others as they join.... however each node has two network interfaces (management & data), and the nodes are joining the cluster with their first interface (mgmt) which is the wrong one.... how can i designate the IP address to use for cluster communication?