1385348 - Never-ending tab throbbers on Nightly with RCWN enabled on Bugzilla and Searchfox

Status: RESOLVED INCOMPLETE

(Core :: Networking, defect)

Description

[Mike Conley (:mconley) (:⚙️)]

There appears to be some kind of RCWN pref flipping experiment underway, and I'm in the cohort that's got it enabled.

I've noticed that since that's happened, I'll visit Bugzilla or Searchfox pages, and the tab loading throbber will never quit, so the page looks like it's loading forever.

Let me know if there's any diagnostic information I can supply - this is pretty easy to reproduce for me.

Comment 1

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

@junior can you provide suggestion to @mconley for further debugging on this issue?

Comment 2

[Junior [inactive]]

ni? related people and block RCWN

Comment 3

[Valentin Gosu [:valentin] (he/him)]

The HTTP logs when reproducing the issue would be very helpful
https://developer.mozilla.org/en-US/docs/Mozilla/Debugging/HTTP_logging#Using_aboutnetworking

Comment 4

[Michal Novotny [:michal]]

I cannot reproduce it. Please, add also cache2:5 to MOZ_LOG.

Comment 5

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

@mconley can you help capture the corresponding MOZ_LOG by following comment #3 and comment #4?

Comment 6

[Junior [inactive]]

I cannot reproduce it, either.

Comment 7

[Mike Conley (:mconley) (:⚙️)]

Test to see if I can reproduce it.

Comment 8

[Mike Conley (:mconley) (:⚙️)]

I've been unable to reproduce it reliably, though I still seem the bug. If it helps, this seems to occur after I've posted a comment to Bugzilla. I tried hammering on https://landfill.bugzilla.org/bugzilla-5.0-branch/ to see if I can reproduce it there, but no luck.

Can I safely disable RCWN in about:config and not disrupt the experiment I'm in? That way, I can see if the problem seems to go away.

Comment 9

[Valentin Gosu [:valentin] (he/him)]

(In reply to Mike Conley (:mconley) - Buried in needinfo / review backlog, eating my way out from comment #8)
> Can I safely disable RCWN in about:config and not disrupt the experiment I'm
> in? That way, I can see if the problem seems to go away.

I think that should be OK. We only gather telemetry when the pref is on.
Although getting the logs for this issue would prove much more useful in debugging it.

Comment 10

[Mike Conley (:mconley) (:⚙️)]

(In reply to Valentin Gosu [:valentin] from comment #9)
> Although getting the logs for this issue would prove much more useful in
> debugging it.

Understood - the problem is that I've had zero luck reproducing it reliably. Let's see if we can at least box it in with the RCWN flip, and work from there.

Setting a needinfo? on myself to report back in a few days with observations on browser behaviour after flipping the pref.

Comment 11

[Mike Conley (:mconley) (:⚙️)]

The problem has not recurred since disabling RCWN. I'll now re-enable RCWN to see if it re-appears again.

Comment 12

[Mike Conley (:mconley) (:⚙️)]

I've had RCWN enabled for a while now, and haven't seen this recur. I'm going to close this as INCOMPLETE. Sorry for the noise, folks.

Comment 13

[Valentin Gosu [:valentin] (he/him)]

Thanks for reporting this. Hopefully it was fixed by something we deployed at some point.