If a client locks some requests and is restarted (migration and so on), these requests won't be passed to the new instance automatically - instead they will stay locked until the locks expires, which is configurable and can possibly take hours.

One possible solution is to use the proposed "List locked requests" endpoint (https://github.com/apify/apify-core/issues/15063) and either unlock them, or just continue processing them as if nothing happened.

• The clientKey changes with every restart -

  crawlee/packages/core/src/storages/request_provider.ts

  Line 34 in 059a33c

  clientKey = cryptoRandomObjectId();

• However, the API should allow clients to work with requests locked by a run with the same ID

Other relevant info:

• https://www.notion.so/apify/Recurring-issues-with-Request-Queue-v2-in-Crawlee-10df39950a2280a693f1cfd449c1194b
• https://apify.slack.com/archives/CD0SF6KD4/p1728384458918029 - previous discussion about the problem
• https://apify.slack.com/archives/CD0SF6KD4/p1729865695638699?thread_ts=1728384458.918029&cid=CD0SF6KD4 - report after the introduction of current implementation