Commit Graph
4 Commits
Author SHA1 Message Date
Daan MeijerandClaude Opus 5.5 0ccb850ea4 Prober: claim ahead of the lanes, not when they run dry
Test and deploy / deploy (push) Successful in 1m34s
Test and deploy / test (push) Successful in 3m6s
The next claim started at half a batch of four pipeline rounds a lane,
3-6 s of work, while a claim takes 1.5-4 s and more; at 20-24 lanes the
lanes waited for candidates up to 19% of their time. queueLevels: the
claim starts below eight rounds of every lane's pipeline (~13 s) and
fills the queue to twice that. Taken candidates go back to waiting at
start-up as before, so a bigger queue costs nothing on a restart.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-28 14:52:48 +01:00
Daan MeijerandClaude Fable 5.1 68b480c97e An "ancient" range a group: articles 1 … low watermark − 1, never filled
Test and deploy / deploy (push) Successful in 3m49s
Test and deploy / test (push) Successful in 3m19s
Every group is tiled from article 1 now: one range with status 'ancient'
covers what had expired on the server before the group was planned.
Nothing selects it (leases, prefetcher, corpus fill, tail merger all take
'planned'), the status panel reports it apart, the search's history stats
leave it out. Migration 6700 adds it for the groups planned before;
Planner.historyRows (now a pure function, unit-tested) for new groups.
The status bands are ordered by an explicit key: the ancient row at
priority 0 tied "history 0-999" with "window 0".

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-28 09:12:05 +01:00
Daan MeijerandClaude Fable 5.1 1810abd7d8 The tail planner grows a group's open range; ranges:merge-tail packs the slivers
Test and deploy / test (push) Successful in 58s
Test and deploy / edge (push) Successful in 2m22s
Test and deploy / deploy (push) Successful in 2m53s
A round added a row a group whenever 100k articles or 15 minutes had passed,
tuned for a tail leased within minutes; with nothing leasing the tail the rows
piled up (127k planned tail ranges of a median 14k articles, 97% adjacent). Now
a round grows the group's last range when it is a planned, unspooled, untaken
row at the tail's priority, up to RANGE_ARTICLES, and only then adds rows; a
range taken meanwhile gets the slice as a row. ranges:merge-tail (--dry-run)
packs adjacent planned, uncaptured, unspooled, untaken tail ranges of a group
and priority into ranges of RANGE_ARTICLES under the lease lock.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-10-27 10:46:47 +01:00
Daan MeijerandClaude Opus 5.5 fd0f2bde75 Rebuild: the live tail
coordinator:run plans the articles after each group's last range every
TAIL_EVERY_SECONDS (120): one LIST ACTIVE per name prefix gives every
group's newest article, and new articles become ranges at priority -1,
leased and prefetched before any history. A group gets a tail range once
it has TAIL_MIN_ARTICLES (100k, a chunk) new or its last planning is
TAIL_MAX_DELAY_MINUTES (15) ago, so quiet groups don't make a range every
round. History (planAll, hourly) is now only for groups not planned yet.
ranges:plan --tail plans one round.

Found on the way: with XFEATURE COMPRESS GZIP on, Eweka compresses LIST
ACTIVE's reply too ("215 NewsGroups Follow [COMPRESS=GZIP]"), and
listActive waited for a plain end that never came (groups:add --top too).
It reads the compressed reply now; the fake server in the test does both.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 10:41:16 +02:00