Closed Bug 421566 Opened 18 years ago Closed 17 years ago

getting Connection Interrupted error pages while using b.m.o

Categories

(Core :: Networking: HTTP, defect, P2)

Product:
Core
Component:
Networking: HTTP
Type:
defect

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: mkmelin, Unassigned)

References

Details

(Keywords: regression)

Attachments

(1 file, 1 obsolete file)

better http log
20.51 KB, text/plain
Details
I'm not sure why, but with the firefox3 betas/trunk I've started getting very frequent "Connection Interrupted" error pages. On firefox2 I don't recall ever having seen one. Now I get them on a daily basis, the site I'm getting it for can be anything from bugzilla to a localhost web application. Sure makes bug 313519 annoying, as sometimes I get them many in a row, making it an effort to view the page. Happens on both windows and linux (and on two different machines, different networks.)
Oh the irony, got it submitting this bug report too...
I'm not getting those... unless my router goes offline then everyone in the house goes off (that is different). Makes me think it's your network and/or router. A lot of things can happen in a network environment. Just all of a sudden problems start occuring that didn't happen before. I had a whole mess of them. I suggest you call your ISP and ask them to assist you and diagnosing your connection and single strength to make sure there are no problems with your line and or equipment. Also I suggest that before calling your ISP you test with Firefox 2 and let us know if you have these problems with Firefox 2. Try for at least half the day or longer. Thanks, Mike
I don't think it's high odds on this being a network/router etc. problem, as I get it both at work and at home on different machines, but more importantly, since I had it the other day for a web application on localhost... It's not reproducible on demand, but I'll keep my eyes open.
i can confirm this also with SM trunk and bugzilla....
I am also seeing this. It occurs randomly, but it seems to happen more frequently when using Bugzilla. ;-) I don't get this at all when using Firefox 2.
Okay well I don't really see it, I think I might have before. Does it say Page not found, Timed out or just Interrupted? Since this seems to be a problem with more than one person I will go ahead and confirm. Thanks, -Mike Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9b5pre) Gecko/2008031109 Minefield/3.0b5pre
Status: UNCONFIRMED → NEW
Ever confirmed: true
this could be related to bug 414477. two questions: 1) do you see this only on https sites/connections? 2) if you go to about:config and set network.http.pipelining.ssl to false, does this go away?
For me it was just Interrupted, but.. I can't seem to get it to happen now, although, it was happening alot during the 1-2 days after installing beta 4. I have to admit now that it may have been a fluke, that just happened to occur on all 3 networks I regularly use (stranger things have been known to happen!).
Flags: blocking1.9?
At least for me I can't reproduce at will, so testing is hard - haven't seen it for a few days now. I had it primarily with bugzilla (so ssl). My case with a web application on localhost was not over ssl - but I guess it's possible that was a problem with the application...
I've been seeing this too with bugzilla, although sporadically. I turned off pipelining for SSL yesterday... Haven't seen happen again, but it's too early to say for sure.
Blocks: 414477
i got this today clicking a link in bugzilla for http://crash-stats.mozilla.com
Needs more info. Is this related to the pipelining, or is this just bugzilla sucking? Reporter: can you try with that preference off and see if it goes away?
Got it again a minute ago with bugzilla, with network.http.pipelining.ssl set to false.
Marking blocking to investigate. Can folks seeing the issues turn off pipelining to see if that helps?
Flags: blocking1.9? → blocking1.9+
Priority: -- → P2
(In reply to comment #14) > Got it again a minute ago with bugzilla, with network.http.pipelining.ssl set > to false. I've also been running with the pref disabled, and just got this error message on bugzilla for the first time in days. So, it would seem that bug 414477 is not strictly the cause of this, although it's possible pipelining makes hitting this problem more frequent... I was getting the bugzilla failure multiple times a day for a while, but since disabling the pref (and the phase of the moon changing), this is the first time I've hit it.
(In reply to comment #16) > this is the first time I've hit it. what if you turn it back on again? ;)
I'm starting to get these inturupted pages now. I didn't used to as I stated above. But very often now I'm getting them and refreshing the page will make it go away.
(In reply to comment #18) to be clear - is this with the pref on, or off; if on, does it still occur with it flipped?
For me: had the pref as default first (true), made it false after filing this bug. So doesn't make any real difference over here. Just got the error page booking a flight, the common thing with bugzilla is that it's one of those pages that take a longer time to load ("searching for suitable flight connections").
would a http log help in this case ? I would try to generate a log if someone really can look into this log.
Yeah, I suspect that would help narrow things down a bit.
Attached file http log (obsolete) —
This is the end of a http log for when I got the page today. Not sure how much of it i should include (and thinderstatus is making it a bit noisy). This was with network.http.pipelining.ssl;false
Attached file better http log —
Bah, this is probably better. The interrupt page happened on process_bug.cgi - should be the one that this log starts with. (Ends a bit after I noticed that tab had an interrupted error page.)
Attachment #311660 - Attachment is obsolete: true
Is this bug happening only on bugzilla.mozilla.org, or does it happen on other sites?
Can we compare the bugzilla server-logs against Magnus' http log to see if a server error occurred at this time?
I see this problem on Safari as well (now listen, I specifically opened safari just to see if I have the same 'connection interruped' problem). Open up a bunch of https://bugzilla... tabs and just reload them a bunch of times. Eventually one will get a 'your connection was interrupted' type error message. Seems to be related to load on bugzilla. I asked Justin F. to investigate, he said he tweaked some knobs on bugzilla and that I should see an improvement. Don't know exactly what he changed. Asking him to comment here...
Re-summarizing -- but not moving to a different component, yet. Would be great to get feedback from Magnus on whether or not it is possible to get this error with Fx3 on a non-b.m.o site.
Summary: getting frequent Connection Interrupted error pages with firefox3 → getting Connection Interrupted error pages while using b.m.o
Also, I should note, I only remember seeing this issue one time since Justin made whatever changes he made, and that's when changing 150 bugs at once.
we improved perf on bugzilla through a few things that would have improved search speed (moved templates from nfs to local disk, etc) but nothing that would have fundamentally fixed this issue for other sites. Can anyone involved list any other sites they see this on? If not, this should be dup'd with the related infra bug.
(In reply to comment #30) > we improved perf on bugzilla through a few things that would have improved > search speed (moved templates from nfs to local disk, etc) but nothing that > would have fundamentally fixed this issue for other sites. Can anyone > involved > list any other sites they see this on? If not, this should be dup'd with > the > related infra bug. > What is that bug #? I've seen this a ton on b.m.o but nowhere else. Haven't seen it recently so fingers crossed that everything is ok..
possible dup of bug 423009 - waiting for other lists here or other feedback in 423009...
(In reply to comment #25) > Is this bug happening only on bugzilla.mozilla.org, or does it happen on other > sites? It's been mostly b.m.o for me, but that may be because of my surfing habits;) Also had it booking a flight via SAS - http://wftc2-sas.e-travel.com
It happeneds on sites site not just Bugzilla, for me.
(In reply to comment #34) > It happeneds on sites site not just Bugzilla, for me. > Ok, where else, and how often?
Just from time to time, any site it's kind of random so it won't help if I give you specific sites (not to mention that I don't have any names right now because I'd have to right it down) -Mike
I haven't seen this since the b.m.o. upgrade - given comment 32 I'm going to take this off the blocker list. If we are sure there is a core issue here please re-nom..
Flags: blocking1.9+ → blocking1.9-
I just got one earlier today.
(In reply to comment #37) > I haven't seen this since the b.m.o. upgrade The b.m.o upgrade had nothing in it that would have helped this issue, afaik.
(In reply to comment #39) > (In reply to comment #37) > > I haven't seen this since the b.m.o. upgrade > > The b.m.o upgrade had nothing in it that would have helped this issue, afaik. > bug 423009 ?
(In reply to comment #40) > bug 423009 ? That's the same bug as this one, it's filed in both places instead of duplicate because nobody seems to be able to figure out who's at fault.
I'm not sure that bug 426550 is indeed a duplicate of this bug, and I have reopened 426550. Relative to comment #20 (and others), the default in SeaMonkey 1.1.9 is network.http.pipelining=False (not True).
(In reply to comment #43) > Relative to comment #20 (and others), the default in SeaMonkey 1.1.9 is > network.http.pipelining=False (not True). the pref we're discussing is network.http.pipelining.ssl, which is newly introduced to trunk and defaults to true for all gecko products. (network.http.pipelining is unaffected.)
I don't think this is a Firefox 3 bug; I've had it happen a bunch of times lately using Firefox 2.
I've not seen this for a long time, WORKFORME maybe?
Yeah, i haven't seen it much myself either lately.
Status: NEW → RESOLVED
Closed: 17 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: