Skip to content

[#369] use correct VIRTUAL_HOST/VIRTUAL_PORT for signald containers

aguestuser requested to merge 369-signald-metrics-not-being-served into main
  • BUG: prometheus unable to scrape metrics from 10 signald containers
  • LIKELY CAUSE: the base yml for all signald containers was providing an (undefined) VIRTUAL_HOST for each container (the old subdomain from when we used to only have one signald p8s server)
  • FIX: provide each signal container with a VIRTUAL_HOST corresponding to its socket id (and refactor configs to group nginx-related configs in prod compose file to make this error less likely)

side-effect:

  • refactor compose files to push all configs that change between prod/dev (such as proxy env vars) to their respective compose files and all values that do not change (such as socket volumes) to the base compose file (even if this is less DRY -- it is less confusing and error prone!)

Closes #369 (closed)

Edited by aguestuser

Merge request reports