Lexy

I worked around it by adding a `pdf_event_hook.py` with the following content: ```python import logging, re, os.path from urllib.parse import urlparse, urlunparse from bs4 import BeautifulSoup def normalize_url(str_url): url =...

I think I now have a better understanding of what is happening. > at point 3 you wite: "The job is started after a while, and we [...]" that make...

Oh my God. `ServerJobCancellationWatcher` is a Hangfire's background process to remove aborted jobs from workers. It frees up the original worker when the job is requeued and at the same...

So, here's what happens. 1. `FetchedJobsWatcher` finds that the job exceeded its invisibility timeout. 2. It removes the job from _Fetched_ jobs and requeues as the next job into the...

> did you fine any clues about how to distingush between the two cases odinserj's SQL storage stores `FetchedAt` time with the `IFetchedJob` instance and terminates the job only if...

I have an idea. What if we store the explicit `Fetched` state in a property of a newly created `RedisFetchedJob` instance, and when the worker calls `RemoveFromQueue`, we compare that...

I managed to make `FetchedJobsWatcher` work as designed, where it cancels and requeues any job that is executing beyond `InvisibilityTimeout`, and it does it consistently in the same session and...

Answering your question, @FixRM. If a job does not support cooperative cancellation with a `CancellationToken`, you can't instruct it to terminate. It'll be holding the worker until the server restarts....

> I'm not sure however about the no. 2. On the servers I manage, I see that if a server restart happens, within a few minutes the jobs that the...

> do you think it could serve it ? Yes, sure. I'd add one thing - requeuing of jobs that belong to reportedly active servers that haven't had a heartbeat...