It stops quite often with Error 502 nginx #53
Comments
tio-trom
changed the title
tio-trom
changed the title
|
Here are more errors I could find: |
|
Can you test this branch? |
|
For sure! Doing it now! |
|
https://ytb.trom.tf/ I got no errors but the instance does not work. Logs here https://paste.yunohost.org/raw/xuhafifudo |
|
Strange... I am able to install and use it. Have you restart NGINX and Invidious services? |
|
Ah ok I had to manually restart those. Works now. Let me test to see if it still keeps on restarting. Thanks! |
|
So far works really well. Didn't interrupt any streams for the past hour, however I still see invidious restarting: But as said no streams stopped for the past hour. It is also that this is a new install of invidious. Could be that once the database is larger and larger, and more users using it, it will start to interrupt the streams again? |
|
Another thing, my system asks to update to Crystal version 1.1.1-1+4.1. Maybe that can break it? I will try not to update it. And, maybe the restart is a scheduled restart for invidious that you may have added in the configs? Maybe that's why it rebooted? |
|
Crystal 1.0.0 was released as a .deb and this is the one we are using in the package. The newer crystall version only provides the .tar.gz package... : / |
|
Ok then. I am going to test with the current crystal version today and tomorrow update to the new one. Just to see if we notice any difference. And oh, ok so there is no scheduled restart process for invidious. Good to know. Wonder why it rebooted then... |
|
I saw again the 502 nginx error: But the instance worked so much better for the past hours since I changed to the branch you recommended. |
|
With the new crystal and all that, I do not think it works well still. It is this dilemma for one it restarts very often again, which is not good, and second if it doesn't restart often then it completely stalls. It will stop working until I restart it manually. |
|
An invidious dev said "the restarts are due to memory leaks or invalid memory accesses (generally, blame QUIC)" - if that helps. |
|
So what I noticed is that it starts to restart more and more often after I did the installation. I suspect the bigger the database, and user base, the more often the restarts. Almost as if our server does not have enough resources...is there a way to increase whatever resources it needs? Because we have plenty of those. At this point our instance restarts ever other minute. This is unusable. And it may also paint a bad picture on yunohost from what I've been talking to the invidious developers who keep on saying it is a yunohost issue and go as far as saying yunohost is only meant for testing stuff. Which I disagreed with, and I do not see this as painting a bad reputation for ynh, but these are the feedback I keep on getting form the invidious devs, at least some of them. I am here to test whatever needs to be tested and even to be a bridge between yunohost invidious and main invidious dev. I you think it may be an invidious issue detail it please so I can bring it to the invidious devs. I hope we can fix this. Thanks! |
|
We discussed this a while ago and the issue seems to lie in Crystal itself, c.f. crystal-lang/crystal#9065 (comment) |
This is from the one who made the comment you are referring to, and she is an invidious dev: " Hello! The goal is to silence the exception raised when the client unexpectedly closes the connection Also, this is not a crash. It's simply an unhandled exception being printed to the logs We (the invidious team) think that a client closing the connection unexpectedly is a normal thing and should be managed by the standard library instead of raising an exception such thing happens e.g when a user has a slow connection or a small drop, and they press the "X" button to stop the page loading. The restarts are due to memory leaks or invalid memory accesses (generally, blame QUIC) I brought to them the issue at hand so they told me this is not the cause. |
|
So it seems invidious restarts quite often, but our yunohost installation restarts so often that it is difficult to use it. |
|
Ok what seems to have fixed it for now is to add to the nginx file for invidious. So far works ok. |
|
please reopen if needed |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The application stops quite often and randomly - I can see this error:
The text was updated successfully, but these errors were encountered: