Background jobs, status: Pending

[Xurbax]

What is the distinction between "Queued" and "Pending" status in the Background Jobs report?

I have Crawler and Inference jobs stuck in the "Pending" state, and have no idea how to fix them...

[Eragos]

Hey!

In my observations means "Queued" not started but awaiting for to be proceed. And "Pending" is started (actually proceed), but not finished.

But it's a bit confusing indeed.

Best Michael

[bmills14]

I'm actually running into a similar/the same issue for inference jobs now as well and am looking for a solution. I mistakenly clicked 'Regenerate AI Tags for Failed Bookmarks Only' prior to linking karakeep to my ollama service. 5 jobs have been stuck in pending for about a week now, with 0 in queued and 0 in failed. I finally got around to linking karakeep and ollama together today, verified by running a Summarize with AI task for one item, then clicked 'Regenerate AI Tags for All Bookmarks'. 145 items were added to the queue, but the original 5 are still stuck in pending and there is no activity in my ollama stack.

So, it seems like this hang up / worker database issue may have been caused by calling a background job for AI with no AI service linked? If so, maybe 'greying out' these options if an AI service is not configured or otherwise stopping the job from being queued in such cases could prevent this hang up from happening to others in the future?

I just started learning how to self host earlier this month and haven't dug into crawlers much yet so can't comment there, but maybe a similar root cause for the initial hang up issue?

[Xurbax]

In my case, the stuck jobs are not due to lack of configured AI service.
I still have some stuck jobs, but since I set every configurable timeout value to quite large values (and even manually changed the monolith hard-coded 5s timeout in the code), I have had few new issues except for the occasional mysterious hang-up in processing for a while. Most (new) jobs have finished (eventually).
Edit: Hmm, after a recent processing stall which as far as I can tell caught up with everything after a while, I have some new stuck Pending entries. I am going to have to take a look at the db I guess and see what these stuck entries are.

[laracabrera]

Has anyone managed to solve the problem of pending jobs? I have a similar issue and there is no error message in the logs.

[Xurbax]

Looking at the queue sqlite db, I have 28 entries just sitting in there. The Background Jobs report says 10 Crawler Jobs and 10 Inference Jobs in Pending. (?? Why the mismatch? Note nothing is actually active at the moment.)
All the db rows say "running", but they are definitely not.

I assume rows should be deleted once the job is complete, right?

[sarbuk]

@Xurbax How did you get at the sqlite db? I'm running the containerized version so not sure how to get to it. Did you resolve the issue in the end, and if so, how? I have several thousand jobs in the queue and pending that I need to sort out (this is from an import):

[mapleshadow]

I have encountered the same problem.

[borgogelli]

How to clear pending job ?

[mrtimothyduong]

I used 'Regenerate Al Tags for All Bookmarks' & 'Regenerate Al Summaries for All Bookmarks' to have AI re-process the bookmarks.

This may not be suitable option for those with a lot of bookmarks, for reference, I have only ~300 bookmarks

[sarbuk]

How to clear pending job ?

Delayed reply I know, apologies... You have to enter the CLI of the container, install the sqlite package using the Alpine package manager, and then edit the database to clear out the jobs that have the pending status. I used AI to help me with the commands.