-
Notifications
You must be signed in to change notification settings - Fork 24
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Remove failed queue items with files. #379
Comments
This is on my ToDo list. Have you mainly noticed the thumbnail task failing when this occurs, or do you see other tasks such as save info or video download? |
The thumbnail task fails and the download video task gets triggered anyway but it stops since there's no stream and stays spinning, never switches to error even though the log says about stream not found. When this happens files in directory are the placeholder thumbnails from twitch (grey image) and a json file(stream info), the VOD registered stays as "processing" and queue item stays as if it was downloading or processing (circle spinning). |
This is the task state after failing : (Adding screenshots so its more clear what i mean hehe) |
I've made some changes in #385 which should catch if the thumbnail download task or the video download task fail in a way that indicates the stream is offline. When this happens it cleans up the workflows, queue item, and vod. I haven't added this logic to the save info task, do you have any information from the API container of how that failed? If you want to test these fixes, build a docker image from this branch #385. Let me know how it works or if it still breaks, if you could provide the debug logs when it happens, that will help. |
Sorry for taking so long to respond, its hard to test because for some reason sometimes it doesn't trigger the error here is the log of the API container exactly when a live archive ended and the second one (the one that fails) gets triggered. The only things i find suspicions are those Also another thing to point out is that the vod that finished correctly and the failed one I noticed that the titles are different VOD ->
|
Update using an image with #385 branch Happened again, the things that are different from past error are:
Workflow panic:
Here is the full log
|
I did not add the cleanup logic to the save info workflow as I was wanting to test if this worked before implementing further. I've updated #385 to cleanup the archive if the task fails in the save info activity. Pull and build the latest version of that branch and give it another try. If/when this error occurs again:
I'll try to recreate this with a low check interval using the same streams in this issue to get some more data. |
With last update in #385 : Both tried to cancelled but got the following error (full log at the end):
3 - Queue item has WorkflowsWorkflow 1 - cb15d1ea-2cc4-4fda-9184-ae1fb0e1cf5b_24 Full log:
|
I setup my demo instance with a channel you had issues with. Everything seems to have worked correctly in my instance. The streamlink download of the video exited with "no playable streams found on this url..." and it went through the cancellation process.
The main temporal Regarding the invalid memory address panic in your logs. I believe that is happen because the temporal client isn't being initialized in the worker. I've updated #385 to setup this up. Can you pull, rebuild, and test again? Also, when you post docker logs can you run |
Can confirm latest #385 commit #f491e16 is working properly and fixes this issue. Just some notes about this:
|
Glad to hear the fixes are working. I'll get this cleaned up and merged sometime this week.
The Twitch API is very unreliable when it updates if the stream is live or not
That makes sense, I send the webhook as soon as the stream is detected as live (which technically it still is according to the API I check). I'd prefer to not have the webhook logic in the workflow and activities as it would then be sent via the worker which I don't feel this right. I don't have a good solution for the double webhook notification right now. |
Fixes released in v2.2.0. |
On a previous issue i brought this up about failed tasks due to the live check going off right after the stream ends and how the only way to mitigate this was increasing the live check interval at the expense of missing X time of a live archive.
Originally posted by @Aerglonus in #360 (comment)
Originally posted by @Zibbp in #360 (comment)
After increasing the interval not by a lot I noticed that even though the task still fails the workers stop properly but the queue item remains registered and failed download thumbnail files remain (since its where it always fails), was wondering if this could be implemented to not having to remove them manually every time it fails and avoid increasing the interval a lot more.
The text was updated successfully, but these errors were encountered: