cannon: Set warp timeout to default 30s from > 1 hour#4994
cannon: Set warp timeout to default 30s from > 1 hour#4994akshaymankar merged 8 commits intodevelopfrom
Conversation
Reduce ping interval to 20s, this is also what most webapps use.
fbb1e1f to
7d5b681
Compare
efd54c5 to
d6ff0ba
Compare
|
The PR description says 20 seconds, but PR title and the changelog file you edited says 30. |
| Set idle timeout on HTTP connections to 30s No newline at end of file | ||
| Set idle timeout on HTTP connections to 30s. Set ping interval to 15s in cannon, | ||
| missing two pings will cause the connection to close. This also removes ability | ||
| for the client to control the ping interval. (#4992, ##) No newline at end of file |
There was a problem hiding this comment.
In this contex, is "the client" anothe backend service, or the frontend? If the latter, we need to double check that no frontend client sets the interval to 30s+, otherwise they will keep be disconnected every 30s. I'm not saying they should do that, I'm saying it will impact them (and they should be fixed, but then should we roll this out)?
There was a problem hiding this comment.
Its the frontend. The webapp doesn't even have a way to use this feature since the browser APIs don't allow sending pings. I remember asking about this few months ago, but we can check again.
The websocket doesn't break if they try to use this feature, they will just not be able to control how often the backend sends pings.
There was a problem hiding this comment.
So it's for clients to ask the backend to send a ping (backend->client) at an interval of their chosing? I had assumed the other way around. Ok this makes more sense.
There was a problem hiding this comment.
Ahh, my wording could definitely be better here. Sorry about the confusion.
Those configs are unused, initially I was going to make all this very extensible, but then I realized this is just a way for us to cause ourselves problems. So I removed it, then I forgot to delete these config options. I will create another PR to clean this up. We're using So the change log is correct. I will update the PR description to match. |
This option was never exposed in the helm chart config, so noone ever configured it. The code using this was removed in #4994
Reduce ping interval to 15s, this is also what most webapps use.
Checklist
changelog.d