Closed Bug 1469015 Opened 7 years ago Closed 7 years ago

Sometimes website pages don't load or load very slowly in Mozilla Firefox Nightly 62.0a1 (2018-06-14)

Categories

(Core :: Networking, defect)

Product:
Core
Component:
Networking
Version:
62 Branch
Platform:
x86_64
Windows 7
Type:
defect
Priority:
Not set
Severity:
major

Tracking

()

Status:
VERIFIED WORKSFORME

People

(Reporter: Virtual, Unassigned)

References

Details

(Keywords: nightly-community)

STR: 1. Open many tabs of website pages and see that some of them won't be loaded or load very slowly I suspect that it regressed in Mozilla Firefox Nightly 62.0a1 (2018-06-14), basing on it, suspected bugs are: - bug #1468503 - bug #1464135 - bug #1466314 Any ideas which could be cause of this issue?
Flags: needinfo?(valentin.gosu)
Flags: needinfo?(amarchesini)
> - bug #1468503 This doesn't seem related. > - bug #1464135 This could be. > - bug #1466314 This could be but only if blobs are involved in POST requests. valentin, are you able to reproduce this issue?
Flags: needinfo?(amarchesini)
I've been getting this issue fairly often for the last 2 days. The sites it happen on are completely random it appears.
Actually, now I think about it, it never happens on the root domain. I'm thinking bug #1468503.
I've tried reproducing it a bunch of times, with no luck. Can you try to track this bug with mozregression? https://mozilla.github.io/mozregression/quickstart.html
Flags: needinfo?(valentin.gosu) → needinfo?(Virtual)
I haven't had time to do mozregression but the bug seems to occur more often when my internet is congested + maybe some packetloss.
Did you have webrender turned on? bug 1467044 seems to have a similar "keeps spinning"/doesn't load issue.
Webrender is on but I never see a spinner, just the page not loading but most of the time it just comes up with SSL too long error. I really doubt Webrender could cause this but I will test with it disabled.
Okay, testing with Webrender disabled I seemed to get the issue less but it still occurs. With Webrender disabled I seemed to get more random long load times but with it enabled I got more of this error after 3secs or so of trying to load: "Secure Connection Failed An error occurred during a connection to www.pokecommunity.com. SSL received a record that exceeded the maximum permissible length. Error code: SSL_ERROR_RX_RECORD_TOO_LONG The page you are trying to view cannot be shown because the authenticity of the received data could not be verified. Please contact the website owners to inform them of this problem." Pokecommunity seems to be the easiest to recreate the issue but it also frequently happened on pages like Google searches and Steam.
So... I just spent ages in Mozregression, silly me forgot to try a really old build first AND realise that the date collides with Windows update kb4284835. It appears this Windows update is the cause but why is it only affecting Nightly and not release is what I'm wondering. I'll need to test more in Mozregression to find the conflict with the Windows update.
Actually, I just updated release to v61 and it is also affected. Ironically this version has some major network changes. I haven't got anymore time to spare but I tested Nightly v61 dated 25/03 and it's not affected.
Thanks for the update. It'll be tricky to verify if the windows update just triggers a bug we have in the code, or if it actively causes the bug. I suspect it's the former, but you never know.
Keywords: qawanted
Okay so I just fully read the 61 patch-notes and they say there's an issue with AVG/Avast and TLS 1.3 (Bug 1468892). I am using AVG IS and did test with it disabled but I assume I forgot to restart the browser. The weird part is I was getting the issue on sites that don't use TLS 1.3 (https://www.pokecommunity.com although it scores a 0/100 on Observatory). Whatever the issue was, it is fixed with the latest Nightly (I checked 1.3 is working fine) with NO AVG update but occurs on old builds. No idea if [:Virtual] was experiencing the same issue as me but my issue looks more like Bug 1468892. As in regards to kb4284835, it was probably just a coincidence but I don't remember an AVG update around that time (AVG reports Jun 18th Build).
Thanks for the thorough investigation Jules. Waiting on :Virtual's ni? before closing this.
See Also: → 1468892
There won't be an Avast software/program update quite yet. However, the first patch for the TLS 1.3 problem from Avast is contained in the AV signature update 180628-08 (see Bug 1468892). That patch rolled out to one of my machines yesterday morning (GMT-7), but didn't show up on another until about 6 hours later.
(In reply to Valentin Gosu [:valentin] from comment #4) > Can you try to track this bug with mozregression? > https://mozilla.github.io/mozregression/quickstart.html I'm very sorry, but now I don't have that much my free time to do that. I don't have WebRender enabled, as it still have too many bugs, especially blurry text fonts. I'm using Avira antivirus without any extensions intercepting HTTPS connections, which in end downgrade my privacy and security. In 2018-06-14 I installed KB4284867, which could also be some kind of cause, like Jules A already wrote and mention this, but this bug doesn't happen in Chromium.
Flags: needinfo?(Virtual)
bug 1468892 comment 60 says avast has patched the problem. Can anyone confirm if that's the case? Reinstalling Avast seems to help in some cases. Otherwise disabling TLS 1.3 might help: set this pref security.tls.version.max;3
Please try to find time to confirm this is fixed. Otherwise we will close this bug for lack of action-ability. Thanks.
Status: NEW → UNCONFIRMED
Ever confirmed: false
Virtual, can you confirm this is fixed?
Flags: needinfo?(Virtual)
I'm still experiencing this issue from time to time.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Flags: needinfo?(Virtual)
(In reply to Virtual_ManPL [:Virtual] - (please needinfo? me - so I will see your comment/reply/question/etc.) from comment #19) > I'm still experiencing this issue from time to time. Comment #16 suggested disabling TLS1.3. Can you try that please (if you haven't already)?
Flags: needinfo?(Virtual)
I'm not able to reproduce this issue anymore for about 2 weeks. Maybe it was fixed by KB4338823 or "It's Magic. I Ain't Gotta Explain..."
Status: NEW → RESOLVED
Has Regression Range: --- → irrelevant
Closed: 7 years ago
status-firefox60: unaffected → ---
status-firefox61: unaffected → ---
status-firefox62: affected → ---
status-firefox63: affected → ---
status-firefox-esr52: unaffected → ---
status-firefox-esr60: unaffected → ---
Flags: needinfo?(Virtual)
Keywords: qawanted, regression
Resolution: --- → WORKSFORME
I like magic. Thanks for the update.
You need to log in before you can comment on or make changes to this bug.