What problem does this solve?
When forge-hw-health starts in periodic (poll) mode, the logs collector drains the full historical log backlog from all BMC log services on the first poll. This floods the analyzer pipeline with stale XID events from faults that may be days or weeks old and already resolved, producing spurious PHRs on every restart.
Feature Description
Add a skip_initial_history: bool flag (default false) to PeriodicLogConfig. When enabled, the collector anchors at the current highest log entry ID on first encounter and skips all historical entries. Only log entries arriving after startup are forwarded.
Code of Conduct
What problem does this solve?
When
forge-hw-healthstarts in periodic (poll) mode, the logs collector drains the full historical log backlog from all BMC log services on the first poll. This floods the analyzer pipeline with stale XID events from faults that may be days or weeks old and already resolved, producing spurious PHRs on every restart.Feature Description
Add a
skip_initial_history: boolflag (defaultfalse) toPeriodicLogConfig. When enabled, the collector anchors at the current highest log entry ID on first encounter and skips all historical entries. Only log entries arriving after startup are forwarded.Code of Conduct