| Summary: | konqueror tasks 100 percent cpu time or pauses or does not finish rendering | ||
|---|---|---|---|
| Product: | [Applications] konqueror | Reporter: | bernhard <e0026053> |
| Component: | general | Assignee: | Konqueror Developers <konq-bugs> |
| Status: | RESOLVED UNMAINTAINED | ||
| Severity: | major | CC: | adawit, andrewb, gassauer, ismail, lux, maarizwan, maksim, treeve01, vkrevs |
| Priority: | NOR | ||
| Version: | 3.5 | ||
| Target Milestone: | --- | ||
| Platform: | Compiled Sources | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
| Attachments: |
snapshot showing the problem
problem rendering http://derstandard.at/ |
||
|
Description
bernhard
2005-10-27 20:29:03 UTC
*** Bug 115285 has been marked as a duplicate of this bug. *** Can't confirm Works fine with konqueror 3.5.1. with 3.5 SVN of today http://orf.at pauses rendering or sometimes does not finish it. (opera displays the page immediately) the rendering pauses/stops at about 80 to 90 %, sometimes it finishes, sometimes not. during this time NO CPU or network activity (gkrellm) is reported. May be this is completely irrelevant: also the kmail-imap connection to my corporate server seems to be "slow" too. is it possible that the address resolution hangs or is inefficient? I frequently get a timeout in konqueror "http://xx.yyy.com" not found, a relaod solves the problem the message I get is "unknown host" I changed it to major, because the curent behaviour will urge me (and others) to abandon konqueror as web-browser due to performance issues (as konqueror was known to be fast) I can't confirm the problem on http://orf.at . Created attachment 15855 [details]
snapshot showing the problem
works fine and fast with firefox
this messages appears if i start konqueror from konsole kio (KIOJob): error 45 tk-proxy2.univie.ac.at hope this helps ERR_UNKNOWN_PROXY_HOST = 45, How can we reproduce this error? I had issues with konqueror-3.4.92 (tasking 100% cpu time...) and reported the bug for version 3.4.92. Since 3.5.2 is out, I could not reproduce it again. I thought, the bug was closed. Thiago - I do not know what happens exactly, at least since today the subjective performance of konqueror is much better. Today I found this (I turned off the proxy) kio (KIOJob): error 14 ad.adworx.at I think there was a problem with the adfilter. Can anyone remember the details? Ferdinand: have you fixed your build? http://bugs.kde.org/show_bug.cgi?id=126333 for a reference. I smell a bogus bug report. No I have no idea how I could fix the build except trying a complete new build, which I have done already. I use kdesvn-build and it should do all the necessary configuration and I do not get any build errors. BTW I have NOT activated the adfilter. Hi, I'd just like to confirm this annoying bug for konqueror 3.5.2. Wikipedia always causes problems, namely konqueror taking up all of the CPU resources. I don't have this problem with firefox. Here's a line from top that I captured. It's only at 98%, but it eventually went up and stayed at 100% for quite some time. 18425 xxx 25 0 59988 32m 20m R 98 6.5 0:17.75 konqueror I too have this problem. Somewhere in the KDE 3.X development something in konqueror has been broken. It feels to be much slower than before. Sometimes not loading pages at all, sometimes taking much time to load the pictures, ... This does not happen with a parallel running firefox on the same machine. It is verified by multiple users on multiple machines. Usually SuSE 10.0 or 10.1 in different states of KDE (3.4, 3.5, 3.5.X, ...). I tried hard to build a test case, but it seems not to be really page specific, but more or less a loading backend problem? Till now I did not succeed to create a test case, but it is really an bug, which forces some to switch to firefox. And I'm annoyed by the slowliness, as I can't switch to firefox due to heavy usage of KIO features in konqueror. Would be nice to get this fixed. Additional note: I thought that is a problem in my overstressed network envirnoment, but maybe it is related to this as well. I often get following error messages (german) "Beim Laden von http://www.heise.de/... ist folgender Fehler aufgetreten: Unbekannter Rechner www.heise.de" (error loading xxx, unknown address xxx). When I load 40 pages from the same server and 5 of these loads present me that message, something must be wrong, as the DNS should be cached already and thus DNS resolution should not fail. I tried the same excessive stress testing with firefox at nearly the same time and no errors occured (but firefox got timing problems with loads of open tabs, where konqueror had none, maybe swapping effects :-). Created attachment 17053 [details] problem rendering http://derstandard.at/ another page with gets rendered only partly. it usually takes a "very long" time to get the missing "header" of this page displayed, firefox renders it "always" immediately. *** Bug 117431 has been marked as a duplicate of this bug. *** *** Bug 121543 has been marked as a duplicate of this bug. *** *** Bug 131779 has been marked as a duplicate of this bug. *** *** Bug 124937 has been marked as a duplicate of this bug. *** *** Bug 133529 has been marked as a duplicate of this bug. *** The page http://www.smh.com.au/ from BUG 133529 is a good example. Really slows down my system a lot. Sigh, looks like this is a catchall bug for performance problems that aren't understood. I just want to point out that on BUG 133529 (www.smh.com.au) it is mainly the X server, not konqueror, that is using high CPU. Yeah, all the merging to this bug is bogus without doing analysis. It seems this is one of this very hard to track down problems, so merging different bug reports may help to find the real trigger for the problem. I tried to be careful with merging. There are some more like this one. If there are bug fixes fixing this problem, but not all the duplicates (or the other way round), then reopening these is required. Splitting the problem "konqueror is slow at lots of pages" into multiple bug reports possibly understates its importance. I could add dozens reports for this, but it's hard to get reproducable test cases. As always, there is pro and contra. Fixing the bug(s) would help. :-) Because bugs have a somewhat similar symptom (with a very loose definition of similar --- hammering on CPU, hammering on X, networks problems are all "slowness" but are extremely unlikely to have anything in common) does not mean they're the same problem. They are not. Some of the things you merged here are easily reproducible and understood, and I am about to commit a fix for one, and another we have a good idea on what roughly is wrong. Not so here. This bug is even worse since the comment #4 and on sounds like a completely different bug than comment #1; a good chunk of the comments seem to be related to proxies, and a good chunk to DNS. I know that. But I also know, that such meta-bugs like "I feel it is too slow" will never be fixed, when one only concentrates on individual aspects of it. I will be more careful with merging to this bug in future and use the "maybe related to" texts. I will be happy if you find and identify individual portions of the problem as individual bugs. Maybe this bug vanishs as well (I hope so, e.g. BUG 133529 fixes look promising to me). But it is a good collection point for the general problem. And allows to show the importance of the problem by voting. P.S. Your last sentence: I do not think it is DNS related, but a caching issue. I found some other older and newer reports which describe different aspects of caching errors. But I have not enough understanding of all the konqueror components to fix anything of this. KDE 3.x is no longer maintained. Feel free to open a new ticket if any of the issues listed here persist in KDE 4.7 or higher. Please do not mix all sorts of bugs into one ticket. |