Announcing signal-sidecar: monitor signal node health

for the last few months, has been using a health and status aggregator for signal nodes (jicofo and prosody) as part of our ingress tooling. it’s now mature enough to open source, and can be found at GitHub - jitsi/signal-sidecar: report aggregated health information from a jitsi signal node

we have a few more features planned that have been captured as github issues, we would love contributions from anyone who finds this tool useful.

from the README:

signal-sidecar collects and aggregates data from a Jitsi signal node for easy consumption by infra tooling. It offers REST endpoints with health, metadata, and metrics. It also provides a HAProxy TCP agent for HAProxy agent-check health checks which allows HAProxy to intelligently balance between Jitsi signal nodes.

Reported drain status is normally based on the contents of a file located at STATUS_PATH. The sidecar will also report DRAIN status in some cases where there is a malfunction.

The HAProxy agent can, using the WEIGHT_PARTICIPANTS flag, send a weight back via the TCP agent that is a function of current jicofo participants vs. PARTICIPANT_MAX. This will never go below 1%.

signal-sidecar is capable of querying the mod_muc_census jitsi-meet prosody plugin to report room census data.