2026.10.8 (#166) added cluster.node-id-reassigned as event 649 without raising TOTAL_EVENT_COUNT, which sizes the collector's per-event level and counter tables. Collector::default() indexed past the end, the inbuxa-collector thread panicked at startup, and the server ran on without writing a line of log or recording a metric.
All three production nodes have been in that state since the 07:20 UTC roll (journal: index out of bounds: the len is 649 but the index is 649). The mail paths were unaffected; only logging and metrics stopped.
TOTAL_EVENT_COUNT 649 → 650.
Two unit tests in trc hold TOTAL_EVENT_COUNT and TOTAL_METRIC_COUNT to the variant lists, check the ids are unique and contiguous, and round-trip to_id/from_id. With the old value the first test fails.
To be released as 2026.10.8.1 and rolled to all three nodes.
2026.10.8 (#166) added `cluster.node-id-reassigned` as event 649 without raising `TOTAL_EVENT_COUNT`, which sizes the collector's per-event level and counter tables. `Collector::default()` indexed past the end, the `inbuxa-collector` thread panicked at startup, and the server ran on without writing a line of log or recording a metric.
All three production nodes have been in that state since the 07:20 UTC roll (journal: `index out of bounds: the len is 649 but the index is 649`). The mail paths were unaffected; only logging and metrics stopped.
- `TOTAL_EVENT_COUNT` 649 → 650.
- Two unit tests in `trc` hold `TOTAL_EVENT_COUNT` and `TOTAL_METRIC_COUNT` to the variant lists, check the ids are unique and contiguous, and round-trip `to_id`/`from_id`. With the old value the first test fails.
To be released as 2026.10.8.1 and rolled to all three nodes.
2026.10.8 added cluster.node-id-reassigned as event 649 without raising
TOTAL_EVENT_COUNT, which sizes the collector's per-event level and
counter tables. Collector::default() indexed past the end, the collector
thread panicked at startup and the server ran on without writing a line
of log or recording a metric.
A unit test now holds TOTAL_EVENT_COUNT and TOTAL_METRIC_COUNT to the
variant lists and the ids to the from_id/to_id round trip.
jcoffey-dev
changed title from Size the collector tables for event 649 to Release 2026.10.8.1: size the collector tables for event 6492026-10-08 17:58:23 +00:00
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
2026.10.8 (#166) added
cluster.node-id-reassignedas event 649 without raisingTOTAL_EVENT_COUNT, which sizes the collector's per-event level and counter tables.Collector::default()indexed past the end, theinbuxa-collectorthread panicked at startup, and the server ran on without writing a line of log or recording a metric.All three production nodes have been in that state since the 07:20 UTC roll (journal:
index out of bounds: the len is 649 but the index is 649). The mail paths were unaffected; only logging and metrics stopped.TOTAL_EVENT_COUNT649 → 650.trcholdTOTAL_EVENT_COUNTandTOTAL_METRIC_COUNTto the variant lists, check the ids are unique and contiguous, and round-tripto_id/from_id. With the old value the first test fails.To be released as 2026.10.8.1 and rolled to all three nodes.
Size the collector tables for event 649to Release 2026.10.8.1: size the collector tables for event 649