987677 - HTTP cache v2 - tests we need help with

Status: RESOLVED WORKSFORME

(Core :: Networking: Cache, defect)

Description

[Honza Bambas (:mayhemer)]

Tracking bug.  

There are currently only 3 devtools and 1 browser/base failing tests we don't know the cause of.

Note: perma-orange bug 948566 is no longer a blocker, disappears when the patch for cache2=on blocking bug 986179 is applied.  However, I slightly believe that other devtools tests could be intermittently failing from the same reason as this one, but it's just a theory.

Please also read behavioral changes of the new cache at (big red boxes) https://developer.mozilla.org/en-US/docs/HTTP_Cache

Comment 1

[Honza Bambas (:mayhemer)]

Panos, would you be willing to provide some help on the bugs being tracked here?  These all are devtools bugs we have no idea (I repeat myself again) how to fix and we would really use some help here.  Seems like dcamp is too busy and you had already some insight feedback on the bugs.

Removing bug 948566 from the list since this one seems to be fixed with a cache2 improvement tracked in bug 986179.

Comment 2

[Panos Astithas (he/him) [:past] (please ni?)]

I can try to look at these 3 tests next week, but I don't know what I have to do to reproduce the failures. Has the new cache landed, but disabled by default? If so, how do I enable it to observe the failures?

I've said it before, but if the only thing blocking the new cache is these 3 tests, then by all means disable them for landing and we will fix them ASAP.

Comment 3

[Honza Bambas (:mayhemer)]

(In reply to Panos Astithas [:past] from comment #2)
> I can try to look at these 3 tests next week, but I don't know what I have
> to do to reproduce the failures. Has the new cache landed, but disabled by
> default? If so, how do I enable it to observe the failures?

Easy: switch "browser.cache.use_new_backend" pref to 1, and you are done.

> 
> I've said it before, but if the only thing blocking the new cache is these 3
> tests, then by all means disable them for landing and we will fix them ASAP.

That would be great, I take your word :)

Comment 4

[Panos Astithas (he/him) [:past] (please ni?)]

I can't seem to get any of these 3 tests to fail on OS X running with the new cache enabled. I've tried running all those tests in both debug and opt builds using --run-until-failure which does 30 runs and I've tried multiple times. Are these failures platform-specific?

Comment 5

[Honza Bambas (:mayhemer)]

(In reply to Panos Astithas [:past] from comment #4)
> I can't seem to get any of these 3 tests to fail on OS X running with the
> new cache enabled. I've tried running all those tests in both debug and opt
> builds using --run-until-failure which does 30 runs and I've tried multiple
> times. Are these failures platform-specific?

Should not be, but each of them has in the title on which platforms it mostly happens.  Maybe they are fixed with recent changes :D  I can try confirm that, but always I just ran 20 or 30 times the same test suite on gum (the project tree we develop cache2 on) it doesn't appear.  Then switching cache2 on on m-i turns stuff orange :/  So hard to say.  Maybe we can try enabling experimentally on m-i again next week or so (currently we miss one important patch to do so now).

Comment 6

[Panos Astithas (he/him) [:past] (please ni?)]

That sounds like the best path forward.