Closed tino closed 1 year ago
The hostname is part of the default metrics.prefix, as documented here:
https://fabiolb.net/ref/metrics.prefix/
This can of course be overridden. The reason for the panic is that your hostname starts with a number, which is an invalid prometheus metric name as documented here:
https://prometheus.io/docs/concepts/data_model/#metric-names-and-labels
Can you verify that defining a metrics.prefix fixes your issue?
Thanks!
I can verify that setting the prefix works!
Thanks
I'm running into panics on startup on rougly 2 out of 10 startups. It fails with:
1ea85cb2f2e1
is the hostname of the container fabio runs in*.Relevant config:
Logs
NB *Why is the hostname prepended? It would be much more logical and useful (IMHO) if that is set as a label. Now I have to apply metric relabelling in prometheus to get aggregated stats (or am I doing something wrong?)