You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
magnetar-admin topic stats — full rate/throughput/size surface:TopicStats now decodes the high-signal PersistentTopicStats metrics it previously dropped: msgRateIn, msgRateOut, msgThroughputIn, msgThroughputOut, averageMsgSize, storageSize, and backlogSize (alongside the existing msgInCounter / bytesInCounter). magnetarctl admin topics stats <topic> emits all of them in its JSON output, so jq '.msgRateIn' (and the out-rate, throughput, and storage/backlog sizes) now work for both non-partitioned and partitioned topics. Fields default to 0 when a broker release omits them. (#293)
magnetarctl message-id output — segmentId no longer dropped: under the scalable-topics feature, topics terminate and topics get-message-id-by-index now surface the PIP-460 segmentId (JSON null when absent) instead of silently omitting it; both commands share one message_id_to_json renderer so their shapes can't drift. (#293)
Changed
CLI (magnetarctl) default log level lowered to warn: the default floor dropped from magnetar=info to magnetar=warn, so magnetarctl is quiet by default and surfaces only degraded-state warnings and errors. The whole -v ladder shifted down one rung — no capability is lost: -v now maps to info (the old default), -vv to debug, -vvv to trace, and -vvvv+ widen into the transport stack (reqwest/hyper/rustls/h2). Scripts that relied on the prior info-level default output must now pass -v. docs/cli.md and docs/logging.md updated to match. (#292)