mirror of
https://github.com/rsyslog/rsyslog.git
synced 2026-06-15 17:32:39 +02:00
Andre Lorbach
389b65fce0
ROSI Collector could get stuck when Loki rejected a pushed sample with HTTP 400, for example because old or clock-skewed sender timestamps were outside Loki's accepted time window. In that state one bad backlog entry could keep the omhttp action retrying instead of advancing to newer logs. Remove the stray /var/log/debug selector from the Loki ruleset. Teach init.sh to hash the installed rsyslog.conf directory before and after copying/generating collector config. If the installed collector rsyslog config changed and an rsyslog Compose container already exists, init.sh recreates only that service with --force-recreate so Docker refreshes the bind-mounted config. First installs still defer to the normal docker compose up -d flow. Document the Loki 400 troubleshooting path, the collector timestamp choice, the automatic init.sh recreate behavior, and the manual checksum check for stale config suspicions. closes: https://github.com/rsyslog/rsyslog/issues/6843