Repository navigation
crawler: 10 s fetch budget, 30-day flag for slow/broken feeds, dead rechecked monthly - #275
Merged
Merged
Conversation
…is not forever Fleet crawler rule (2026-11-08): - CRAWL_FETCH_TIMEOUT_MS = 10_000 in packages/feed/src/fetch.js replaces the 15 s literal; the deadline now starts before DNS so lookup + connect + headers + body share one budget. - Thrown fetch errors are classified (timeout, dns-failed, tls-failed, connection-failed, fetch-failed); a name that does not resolve is dns-failed, no longer blocked-host. EAI_AGAIN is dns-retry and keeps the short ladder so a resolver hiccup cannot park the directory for a month. - crawlFeed: timeout / DNS / TLS / connection / 5xx failures set next_fetch_at = now + 30 days (last_error, error_count recorded, feed interval kept). 4xx and no-feed-found keep the 1h..1d ladder. - 'dead' (10 failures) is rechecked every 30 days: recheckFeeds() reads status='dead' rows past next_fetch_at (new partial index feeds_recheck_idx), dueFeeds offers them a 5% share plus any slack, a success restores the feed to active. Migration re-stamps existing dead feeds to last_fetched_at + 30 days. - 429 / 503+Retry-After handling unchanged (capped at a day). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ralyodio
added a commit
that referenced
this pull request
Nov 8, 2026
Follow-up to #275. 721 dead feeds were retired by older rules with fewer than ten failures (garbage URLs, not-a-hostname); a failed recheck recomputed status from error_count and downgraded them to 'error', putting them back in listings and on the 1h..1d ladder. markCrawlFailure now keeps 'dead', and crawlFeed waits 30 days for any dead feed (status is in DUE_COLUMNS). Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Implements the fleet crawler rule (2026-11-08) for the feed poller.
CRAWL_FETCH_TIMEOUT_MS = 10_000(was 15 s) covering DNS + connect + headers + body; thrown errors classified intotimeout/dns-failed/tls-failed/connection-failed/fetch-failed(dns-retryfor EAI_AGAIN stays on the short ladder).next_fetch_at = now + 30d(last_error + error_count recorded, the feed's interval kept). 4xx / no-feed-found keep the 1h..1d ladder.deadis no longer forever:recheckFeeds()(partial indexfeeds_recheck_idx) hands due dead feeds todueFeeds(5% share + slack); one recheck per 30 days, success restoresactive. Migration re-stamps existing dead feeds tolast_fetched_at + 30d.Tests:
packages/feed/test/crawl-timeout.test.js,packages/ingest/test/recheck.test.js, recrawl + db tests updated. Full suite green locally on Node 22.🤖 Generated with Claude Code