- Batches that fail go to /data/spool (SPOOL_MAX_MB, 1 GiB by default) and
are sent again oldest first; retries no longer block the store loop and
follow the shutdown context.
- Docker and host logs wait for room in a full queue instead of being
dropped; the Docker position only moves once a line is stored or spooled.
- Reverse DNS no longer holds up the syslog listeners, with an LRU cache
and a cap on concurrent lookups.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
The histogram above the list becomes a configurable timeline:
- Scale linear / sqrt (default) / log, height S/M/L, bars or area.
- Colors: stacked by severity (error+, warning, the rest), intensity
against the median of the window (calm, burst >3x, anomaly >10x), or
none; colors are CSS variables with light and dark values.
- Vertical graduations on round local times (hh:mm:ss, hh:mm, dd/mm).
- Tooltip: interval bounds, total and detail per severity.
- Division automatic (~100 intervals) or fixed (1 s to 1 day), capped at
300 intervals by the server; intervals aligned on the local time zone.
- Refresh off / 5 s / 15 s / 30 s / 1 min / at each new interval. In
live mode the last interval is incremented from the SSE stream and the
timeline reloads at each new interval; paused while the tab is hidden.
- Click a bar to zoom on its interval, drag to zoom on a selection: the
list, export and counters follow through new from/to parameters.
The timeline now runs on the server clock (bounds and "now" come from
/api/histogram): the axis was drawn from the browser clock and refreshed
every 30 s only, so a clock difference with the server or a hidden tab
left it behind the logs.
/api/histogram returns interval indexes with the count per severity; the
division logic lives in histogram.go with table-driven tests.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
cedricandClaude Opus 5.5