Cannot publish a large channel

[pcenov]

Observed behavior

Observed while testing #4002 today - every time I attempt to publish a large channel (5K resources) at some point it seems to time out without any clear indication to the user. Sometimes both the Unpublished and Publish buttons are displayed simultaneously.

Expected behavior

The channel should be published successfully. In case there is a server timeout or any other type of error this should be communicated to the user as otherwise it's quite confusing.

Steps to reproduce the issue

1. Go to https://hotfixes.studio.learningequality.org/en/accounts/#/ and sign in
2. Create a brand new channel add import multiple folders and resources from a large channel such as CK-12
3. Click the Publish button and wait

Additional information

Video:

2023-06-06_15-18-46.mp4

Usage Details

• OS: Windows 10, Ubuntu
• Browser: Chrome
• URL: https://hotfixes.studio.learningequality.org/

[pcenov]

@bjester, @radinamatic

[bjester]

@pcenov Could you import resources from a different channel, and try again? CK-12 has some issues which don't popup on production, but do on hotfixes.

[pcenov]

@bjester sure, I thought it was server related. Are all other channels a good game or there are some which should be avoided as well for this particular scenario?

[bjester]

@pcenov This is the only problematic channel that I recall, so others should be okay

[pcenov]

Hi @bjester, yesterday I was able to publish a channel with more than 2K resources.
Today I tried with the following channel 9139ab6e91054da9bb708ea84fde3439 which I created by importing resources from Khan Academy US (about 5500 resources). So after about an hour and a half it had reached about 45% and then when I checked again it was showing Publishing channel 0%:

I was testing in Chrome with the memory saver off. I tried refreshing the page and signing out and signing in but nothing changed. Then after about an hour when I looked again the status had changed to published.
So I confirm that publishing a large channel with resources which are not from CK-12 is working but my guess is that there are issues with the hotfixes server which are resulting in the above strange behavior.

[bjester]

So I believe the issue @pcenov has found is the recurring 0% issue. I can only suspect that the task ID is getting wiped out after a certain amount of time, perhaps based of our fetch expiration, which sounds reasonable if the channel is large and takes a while to publish.

[vkWeb]

@bjester didn't we fixed a similar issue while back with celery's late acknowledgement? 🤔

[bjester]

@vkWeb I think the two issues likely produced similar symptoms, since they both involve lengthy publishing operations, but I don't believe this UI issue is directly caused by task issues

[vkWeb]

@bjester okay!