Skip to content

crawler: tunable site-start concurrency; frontier pool 3 per process (0.49.2) - #196

Merged
ralyodio merged 1 commit into
mainfrom
pitbot-site-concurrency
Nov 8, 2026
Merged

ralyodio merged 1 commit into
mainfrom
pitbot-site-concurrency

Conversation

@ralyodio

@ralyodio ralyodio commented Nov 8, 2026

Copy link
Copy Markdown
Contributor

Two ceilings hit while scaling toward 1M pages/h:

  • Site starts: they ran max(2, concurrency/4) = 8 at once. A start is mostly politeness waits (robots, sitemaps, home page and probes, 1.5 s apart), so too few sites were active: 144k page jobs were delayed and 0 were ready. CRAWL_SITE_CONCURRENCY now sets it, and dev2 will run 64.
  • Frontier connections: each process opened up to 8 Postgres connections. 11 processes held 70 idle connections, the pitcrawl role's cap, so the 12th could not connect and its site starts hung. CRAWL_DATABASE_POOL now sets it, defaulting to 3.

Crawler: 131 tests pass.

🤖 Generated with Claude Code

…3 (0.49.2)

Two ceilings hit scaling toward 1M pages/h (2026-11-08):
- Site starts ran max(2, concurrency/4) = 8 at once on dev2. A start is mostly
  politeness waits (robots, sitemaps, home page, probes, 1.5 s apart), so too
  few sites were active: 144k page jobs delayed, 0 ready. CRAWL_SITE_CONCURRENCY.
- Each process opened up to 8 frontier connections; 11 processes held 70 idle,
  the pitcrawl role's cap, and the 12th could not connect, so its site starts
  hung. CRAWL_DATABASE_POOL, default 3.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@ralyodio
ralyodio merged commit 85fd577 into main Nov 8, 2026
4 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant