From c73e9ce30c28f8be2ebbdc8895096e317f62070b Mon Sep 17 00:00:00 2001 From: Ilya Kreymer Date: Fri, 19 Dec 2025 22:18:01 -0800 Subject: [PATCH] remove extra sleep --- src/indexer.ts | 4 ---- src/util/state.ts | 2 +- 2 files changed, 1 insertion(+), 5 deletions(-) diff --git a/src/indexer.ts b/src/indexer.ts index 6a466e97..993ef23c 100644 --- a/src/indexer.ts +++ b/src/indexer.ts @@ -9,7 +9,6 @@ import { initRedisWaitForSuccess } from "./util/redis.js"; import { AsyncIterReader } from "warcio"; import { RedisDedupeIndex } from "./util/state.js"; import { basename } from "node:path"; -import { sleep } from "./util/timing.js"; export type DedupeIndexEntry = { name: string; @@ -138,7 +137,6 @@ export class CrawlIndexer { } logger.info("Done!"); - await sleep(30); await dedupeIndex.markImportFinishedTS(); process.exit(ExitCodes.Success); } @@ -198,7 +196,6 @@ export class CrawlIndexer { if (res && res.size) { await dedupeIndex.addStats(res.size - size, crawlId, commitToAllkey); } else { - console.log("NO DUPE", hash, res); await dedupeIndex.addRevisitSize(hash, size, crawlId); } } else if (url && date && hash) { @@ -210,7 +207,6 @@ export class CrawlIndexer { crawlId, commitToAllkey, ); - console.log("MATCH DUPE", hash, size); await dedupeIndex.matchRevisitSize(hash, size, crawlId, commitToAllkey); } else { logger.warn("Skipping invalid CDXJ, data missing", { diff --git a/src/util/state.ts b/src/util/state.ts index 8b6dc04a..ed23db4c 100644 --- a/src/util/state.ts +++ b/src/util/state.ts @@ -401,7 +401,7 @@ export class RedisDedupeIndex { const { size, crawlId } = JSON.parse(res); await this.addStats(origSize - size, crawlId, commitToAllKey); } catch (e) { - console.log(e); + logger.debug("Error adding revisit size", e, "state"); // ignore } }