maverick/search-engine
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Crawler: shard-aware seeding for distributed crawling

Browse source 
Add SHARD_ID/SHARD_COUNT env: each worker seeds only domains where
hash(domain) % SHARD_COUNT === SHARD_ID (FNV-1a). Defaults to single-node
(all domains). Crawler only follows same-domain links, so shards never
overlap and lose no coverage vs a single crawler.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
King Omar 2026-07-22 23:17:35 +10:00
parent 9d3be71469
commit 7fa8114a7f
1 changed files with 15 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

17
crawler/src/index.js
View file

@ -14,6 +14,18 @@ const MAX_DEPTH = parseInt(process.env.MAX_DEPTH || '2') // link
const MAX_PAGES_PER_DOMAIN = parseInt(process.env.MAX_PAGES_PER_DOMAIN || '30') const MAX_PAGES_PER_DOMAIN = parseInt(process.env.MAX_PAGES_PER_DOMAIN || '30')
const HOST_DELAY_MS = parseInt(process.env.HOST_DELAY_MS || '1000') // politeness per host const HOST_DELAY_MS = parseInt(process.env.HOST_DELAY_MS || '1000') // politeness per host
const FRESH = process.env.FRESH === '1' const FRESH = process.env.FRESH === '1'
// Distributed crawling: each worker owns a disjoint slice of the seed domains
// (hash(domain) % SHARD_COUNT === SHARD_ID). Defaults = single-node (all domains).
const SHARD_ID = parseInt(process.env.SHARD_ID || '0')
const SHARD_COUNT = parseInt(process.env.SHARD_COUNT || '1')
// FNV-1a — cheap, stable hash so every worker agrees on who owns a domain.
function hashDomain(s) {
let h = 0x811c9dc5
for (let i = 0; i < s.length; i++) { h ^= s.charCodeAt(i); h = Math.imul(h, 0x01000193) }
return h >>> 0
}
const ownsDomain = domain => SHARD_COUNT <= 1 || hashDomain(domain) % SHARD_COUNT === SHARD_ID
const sleep = ms => new Promise(r => setTimeout(r, ms)) const sleep = ms => new Promise(r => setTimeout(r, ms))
const lastHit = new Map() // host -> last fetch ts (politeness) const lastHit = new Map() // host -> last fetch ts (politeness)
@ -24,12 +36,13 @@ function stats() {
} }
function seed() { function seed() {
let n = 0 let n = 0, skipped = 0
for (const { rank, domain } of readDomains(SEED_DOMAINS)) { for (const { rank, domain } of readDomains(SEED_DOMAINS)) {
if (!ownsDomain(regDomain(domain))) { skipped++; continue }
F.enqueue(`https://${domain}`, regDomain(domain), 0, rank) F.enqueue(`https://${domain}`, regDomain(domain), 0, rank)
n++ n++
} }
console.log(`Seeded ${n} domains`) console.log(`Seeded ${n} domains (shard ${SHARD_ID}/${SHARD_COUNT}, skipped ${skipped} out-of-shard)`)
} }
async function politeWait(host) { async function politeWait(host) {