1.5.0.6 and "Server not found" is still with us

Discussion of bugs in Mozilla Firefox
Locked
alta88
Posts: 1029
Joined: January 28th, 2006, 3:08 pm

Post by alta88 »

trolly wrote: or a windows update which was scheduled at the same time is the problem. Don't know what it could be.


sorry that's fairly ridiculous straw-grasping.

this is a Fx bug. period. why aren't you recommending that the OP, who did an excellent job documenting the scenarios, file a bugzilla report? and then have people, by swarms, vote for it?
User avatar
malliz
Folder@Home
Posts: 43796
Joined: December 7th, 2002, 4:34 am
Location: Australia

Post by malliz »

Sorry but can you prove what you say?^ I can think of any number of things that could go wrong with dial-up modems and timing, put up or shut up
What sort of man would put a known criminal in charge of a major branch of government? Apart from, say, the average voter.
"Terry Pratchett"
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

The problem is that this problem only affects a special group of people. I can not recommend filing a bug when the cause is not clear. That bug will either hang indefinitely or closed as invalid because there is no way to reproduce the problem on other machines. It seems like a all-or-nothing problem. Either you have it or you have not and never will. In this case no dev will dig into this. Address resolution is an OS feature and FF simply asks the OS to do it. Maybe IE simply retries a few times or take the page silently from cache if this problem shows up but FF will display an error. I have broadband for years and i never ever have or had this problem starting from FF 1.0.x over 1.5 to trunk on W2K, WXP and Linux. And i think that is also valid for the majority of FF users.
So the chance that it is accepted as bug is negligible. If you can tell that this or that OS setting is affecting this this is another problem but until then ...
Think for yourself. Otherwise you have to believe what other people tell you.
A society based on individualism is an oxymoron. || Freedom is at first the freedom to starve.
Constitution says: One man, one vote. Supreme court says: One dollar, one vote.
User avatar
malliz
Folder@Home
Posts: 43796
Joined: December 7th, 2002, 4:34 am
Location: Australia

Post by malliz »

Always been on dial-up and I have never had this problem even with the 100's of nightlies I have tested over the years. And my phone lines are old and I am quite some way from an exchange (2 kilometres). It would help if we could say all the problems are coming with the 98 & Me OS's or a particular AV program, or even one or two ISP's but we cannot. One thing I would try personally if I was having dial-up problems would be to drop the connection speed of the modem down a notch.
What sort of man would put a known criminal in charge of a major branch of government? Apart from, say, the average voter.
"Terry Pratchett"
ApTeM
Posts: 89
Joined: September 3rd, 2006, 1:32 am
Location: Artem S. Tashkinov

Post by ApTeM »

I have never ever had this problem (and so did most my friends who use FF) so people, please, stop whining and set up your network connection properly. Don't play too much with firewalls and other things you don't fully understand.
ApTeM
Posts: 89
Joined: September 3rd, 2006, 1:32 am
Location: Artem S. Tashkinov

Post by ApTeM »

Don't complain about your ISPs in this forum. It's about Firefox, not about your cables/hi-speed modems/etc.
alta88
Posts: 1029
Joined: January 28th, 2006, 3:08 pm

Post by alta88 »

trolly wrote:The problem is that this problem only affects a special group of people. I can not recommend filing a bug when the cause is not clear. That bug will either hang indefinitely or closed as invalid because there is no way to reproduce the problem on other machines. It seems like a all-or-nothing problem. Either you have it or you have not and never will. In this case no dev will dig into this. Address resolution is an OS feature and FF simply asks the OS to do it. Maybe IE simply retries a few times or take the page silently from cache if this problem shows up but FF will display an error. I have broadband for years and i never ever have or had this problem starting from FF 1.0.x over 1.5 to trunk on W2K, WXP and Linux. And i think that is also valid for the majority of FF users.
So the chance that it is accepted as bug is negligible. If you can tell that this or that OS setting is affecting this this is another problem but until then ...


i can't disagree more with almost everything you've said. not filing a bug because it affects 'special group' or 'cause is not clear' is nonsensical. are you interested in improving Fx quality or sweeping things under the rug?

this is a difficult bug that won't be solved by the 1st level advice given here. it will require a skilled developer or QA person with knowledge of how to use a debugger and step through the Fx code to examine storage and buffers as the error is thrown. (and that's c/c++ code btw, not javascript..)

the OP's problem has been gone through far better here than most problem reports; there is *plenty* of supporting info to file a bug. have you or the other 'advisers' even checked if one already exists?

probability of whether a Fx dev sees the bug or addresses it is guaranteed to be zero if the bug is not filed. the OP and others could recreate this regularly, reported on all kinds of networks connections. if it works in IE for the reason you hypothesize, then it is *still* an Fx bug for not handling the error. and there are reports of no other change than Fx version to bring this out.

whether it WFY or others who chime in doesn't negate the error, or do you think it does? you are closing the feedback loop to Fx devs with your advice - this is a huge disservice to the Fx big picture.
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

You are free to file a bug. There are hundreds of bugs which are still at UNCONFIRMED because no one is seeing it too or being able to reproduce it or providing the necessary informations to find the problem. Simply saying that it does not work will lead only to a patch not a fix because you can cure the symptoms (with e.g. trying two or three times before reporting an error) but this will not really fix the problem.

If you say some one is needed who can operate a debugger and more you are probably wrong. The person which should be able to fix it must see the problem. Otherwise it is impossible to do. And - assuming it is a timing problem - using a debugger will probably cover the problem because the timing is different inside a debugger. Programming is my job for almost 20 years now and i know what i'm talking about.

BTW, i have seen reports that IE simply displays a cached page instead of reporting an error when the server is not found.
VanillaMozilla
Posts: 13808
Joined: November 7th, 2005, 11:26 am

Post by VanillaMozilla »

Alta88,
This forum is littered with people who jumped to conclusions and then had to recant. In the case of the original poster, we know that his computer has problems fetching IP addresses, and IE had a problem. That's not good material for a bug report.
Last edited by VanillaMozilla on September 15th, 2006, 9:08 pm, edited 1 time in total.
VanillaMozilla
Posts: 13808
Joined: November 7th, 2005, 11:26 am

Post by VanillaMozilla »

Speaking of the original poster if we can get back to his problem,

@IDon'tKnow:
Dickvl was asking if you can access the sites by IP address. At some time when 30 or 40% of sites fail to open, the question is whether you can access the sites if you use the IP address instead of the URL. That's an important question because it helps prove whether the problem is domain-name lookup.
alta88
Posts: 1029
Joined: January 28th, 2006, 3:08 pm

Post by alta88 »

trolly wrote:If you say some one is needed who can operate a debugger and more you are probably wrong. The person which should be able to fix it must see the problem. Otherwise it is impossible to do. And - assuming it is a timing problem - using a debugger will probably cover the problem because the timing is different inside a debugger. Programming is my job for almost 20 years now and i know what i'm talking about.


i don't know how you can be a programmer and not use a debugger on this. yes, it is obvious the error must happen when using the debugger.

however, i think you are correct that this might be a timing problem, one of the most difficult to debug in a multitasking app in general, and specifically in networked apps. i'd guess one line of tight retry loop code in the right place would fix this for questionable networks. so the question is: what's the right process to get it to a Fx dev???

in the meantime perhaps the OP could try this.

look, all i want is legitimate bugs taken seriously and to get fixed. there are too many posters here, with bad "it ain't Fx" attitudes, like the ones above, who blow off bug reporters. it takes some skill to know among user error, config fix, Fx bug; those who can't tell shouldn't post.
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

A debugger will not help here because DNS is an Operating System Service which means that the code will jump into the kernel space where you can not follow.

You may check the Windows call GetAddressByName ( http://msdn.microsoft.com/library/defau ... name_2.asp ). There is no timeout parameter.

Once Vista is out maybe this call GetAddrInfoEx ( http://msdn.microsoft.com/library/defau ... infoex.asp ) is better because it provides a timeout. But it is only supported in Vista.

Most probably this call GetAddrInfo ( http://msdn.microsoft.com/library/defau ... info_2.asp ) is used (available on all Windows OS releases) but it does not provide a timeout too.

So i do not see anything where the FF devs can tune.

You can either file a bug or try to connect a dev using their addresses published in the bugs at bugzilla but i do not know if it helps. To get it fixed you must have
a) a machine where it happens every time or often
b) a dev which has access to such a machine
c) an idea what is wrong not what symptoms you see

BTW: Many posters is a subjective term if compared to over 100 million downloads. I do not doubt that there is a problem but i have no idea which one.
alta88
Posts: 1029
Joined: January 28th, 2006, 3:08 pm

Post by alta88 »

trolly wrote:A debugger will not help here because DNS is an Operating System Service which means that the code will jump into the kernel space where you can not follow.


no. any caller of the MS api will have to set up storage per the call's spec; if the call returns, what is in that storage is clues. if the call doesn't return, that's a clue. i expect Fx then calls the error routine, which can return thousands of errors, quite specific. inspecting all this before throwing up the Not Found message is the 1st step. do you know if debug verbosity logging exists in this app?

trolly wrote:So i do not see anything where the FF devs can tune.


very incorrect. do you have experience designing network socket level apps? any programmer who claimed it was the snow and not code causing transient errors in the satellite networked clients app, would be fired.

here's a diagram for you:
1. call returns 0
-> great.
2. call call returns non-zero
-> check the RC. it could contain status that means permanent error explicitly
-> retry if non permanent error!
-> then throw a verbose, meaningful, helpful error message.
3. call never returns (user sees hang)
-> call must be set up on timer that pops after certain time if no answer
-> retry
-> throw 'no answer' verbose message

trolly wrote:You can either file a bug


not appropriate for me to do it, as it doesn't happen to me. the OP, who seemed quite savvy, is the appropriate person, but i'm sure is long gone after several pages of useless advice. frankly, a simple examination of the Fx code leading up to the throwing of the error will tell you everything you need about whether the design is solid in handling transient network exceptions.

trolly wrote:BTW: Many posters is a subjective term if compared to over 100 million downloads. I do not doubt that there is a problem but i have no idea which one.


i don't understand what this means. let me be clear on what i meant: the driveby troglodite Fx bully is just a fact of life; when mods or wannabe jr mods or anyone with an officious mozillazine tag belittle bug reporters, offer bad advice, blame it on 'windows download', refuse to file bugs, or simply have no clue about software, then that's just BS. and extremely destructive to Fx.

you have no idea (nor do i) how many people are out there, silently hitting refresh, constantly. but apparently it's enough of a problem for someone that they wrote an extension to Retry.

think i'm finished with this one..
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

Well, i'm building networks apps for a few years now. But they are mostly running in a very special environment so i never had the problem thinking about DNS.
If i have time - and my time is very limited - i may dig into the code and find out what FF is doing here. (just downloading the 1.5.0.7 code)

You're probably right about the numbers but it seems that most people does not regard it as a bug.
VanillaMozilla
Posts: 13808
Joined: November 7th, 2005, 11:26 am

Post by VanillaMozilla »

Gentlemen,
Have we forgotten the original poster?

Alta,
Since you have a clear concept how to do this, you should really just have a try. You really can't give directions in a user support forum on how to debug a complex application. We're just users here. And you really couldn't expect much more, even if this were a technical forum. This just isn't the place.
Locked