#4163 Endless polling with errors on PR pages to port :8088
Closed: Invalid by pingou. Opened by techtonik.

If you open any PR page on this site and open console in Chrome, there is an endless stream of failed connection attempts to port :8088.

GET https://pagure.io:8088/pagure/pull-request/4152 net::ERR_CONNECTION_RESET

This is harmless, it's the event source server that is not accessible, it can be entirely ignored.

Note that this should also happen on issues

(I also do not have this behaviour on firefox, could you check?)

Firefix fails with "CORS did not succeed" only once.

Why event source server is not important? Is there a status page for pagure services?

Why event source server is not important? Is there a status page for pagure services?

It's not that it is not important, but pagure's code handle being deployed without it and if the server is down or not accessible (as is the case right now, at least until I restarted stunnel) pagure's behaviour will be mostly unchanged.
The EV server is de facto optional when deploying pagure.

Opera returns a ping string over time when not being accessed through a proxy such as squid

====================================

FF returns:

https://pagure.io:8088/pagure/pull-request/4152

Problem loading page

The proxy server is refusing connections

Firefox is configured to use a proxy server that is refusing connections.

Check the proxy settings to make sure that they are correct.
Contact your network administrator to make sure the proxy server is working.

because of course the squid proxy is not proxying content on non-standard ports in this case

But why bother the end user with the ping across port :8888 in the first place ... ?

But why bother the end user with the ping across port :8888 in the first place ... ?

The port is 8088 and that's where the UI connect for server-sent event, basically what allows pagure to update the UI without reloading the entire page. It's a read-only light-weight and http-based equivalent to websocket.

I'm inclined to close this ticket if I've answered all of your questions. stunnel which gets sometime "too many connection" has been restarted so the symptom for this has been fixed and this situation should not affect pagure's behavior.

Metadata Update from @pingou:
- Issue close_status updated to: Invalid
- Issue status updated to: Closed (was: Open)

Metadata