1.5.0.6 and "Server not found" is still with us

Discussion of bugs in Mozilla Firefox
Locked
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

Just to close it:
prnetdb.c lines 2133 ff

Code: Select all

        rv = GETADDRINFO(hostname, NULL, &hints, &res);
        if (rv == 0)
            return (PRAddrInfo *) res;

        PR_SetError(PR_DIRECTORY_LOOKUP_ERROR, rv);
A bit little error checking here. And definitely no checks for the temporary error WSATRY_AGAIN. But it is in a multiplatform part.

If anyone wants to try this:

Code: Select all

#ifdef WIN32
        for (int i = 0; i < 3; ++i)
        {
#endif
        rv = GETADDRINFO(hostname, NULL, &hints, &res);
#ifdef WIN32
        if (rv == WSATRY_AGAIN)
        {
          Sleep(100);
          continue;
        }
        else
#endif
        if (rv == 0)
            return (PRAddrInfo *) res;
#ifdef WIN32
        else
          break;
        }
#endif
        PR_SetError(PR_DIRECTORY_LOOKUP_ERROR, rv);
It will repeat the resolution three times if there is a temporary error reported.

I will file a bug for that.

Edit: Added to this bug: https://bugzilla.mozilla.org/show_bug.cgi?id=337243
Feel free to vote for it. If i can manage to build it with this patch i can offer it for tests.
Last edited by trolly on September 18th, 2006, 4:15 am, edited 1 time in total.
fazza
Posts: 3
Joined: September 18th, 2006, 3:56 am

Post by fazza »

im having the same problem since i updated from 1.5 to 1.5.7 but i still get the same error now that i have uninstalled 1.5.7 and gone back to 1.5 i even reformatted prior to this because i thought i had a virus or such but i think it is a bug in fx
fazza
Posts: 3
Joined: September 18th, 2006, 3:56 am

Post by fazza »

sorry that would be the ( cant find server ) error
VanillaMozilla
Posts: 13808
Joined: November 7th, 2005, 11:26 am

Post by VanillaMozilla »

EDIT: Trolly, after actually looking at your code example, it looks like you have actually found something. Good work.

It helps to know that there is a bug. However, I hope people do not just dismiss these complaints as just a bug in Firefox. We know that many or most of the problems users see are due partly to a breakdown in their network environment, and there are still steps that can be taken to reduce or eliminate the problem. The notorious difficulty in persuading users to fix their firewalls is ample demonstration that users need encouragement to troubleshoot. And networking problems are even harder to find.


<del>There have been a few examples in which I wondered if other browsers were more fault-tolerant, and it wouldn't hurt for someone who is knowledgeable about networking to have a careful look at all the links in the chain. Nevertheless, we know that many problems are due to undiagnosed networking problems. Some people have reported that the problem was solved either with TreeWalk or by specifying a known, reliable DNS service.

For this reason, I believe it would be a profound disservice to most users to just dismiss their complaints as bugs in Firefox. They need to be encouraged instead to take all measures to make sure their networking environment is working correctly. The notorious difficulty in persuading users to fix their firewalls is ample demonstration that users need encouragement to troubleshoot. And networking problems are even harder to find.

There are some specific problems with ascribing support requests to bugs. First, in cases where the user can be persuaded to actually persevere in finding the problem, the problem is often traced to a networking problem or some other problem other than Firefox. Most users either solve the problem or give up before they get that far. Unfortunately, in many instances, it's not feasible to solve the problem because the user does not give correct information or has difficulty in following the instructions (which may also be confused and conflicting).

Second, we see this with every release. A common assertion is that version n doesn't work while version n-1 does, and therefore <i>there must be something wrong with version n</i>. While I don't doubt that a problem occurred suddenly, even perhaps after an update, <i>it cannot be true for all n that version n is flawed while version n-1 is not.</i> So the reasoning on this argument is certainly incorrect.

There are assertions that other browsers work while Fx fails. At least some of these reports appear to be incorrect. A typical report is that Fx failed and then IE succeeded. On the other hand, there are also reports that Fx failed but succeeded on the second try. When I asked a user to try IE for a while, he found that IE had a problem too, but he hadn't noticed. Apparently the first attempt fails while the second attempt succeeds. On this forum the first attempt is usually with Fx.

On the other hand, there is one case in which Fx reportedly failed to open random sites, while Opera suceeded (slowly) with a different set of random sites, so maybe there is a difference. Since Fx works fine in the vast majority of cases, it seems like any problem would have to be in handling of faults in the network environment.

Is it possible to improve on fault tolerance? The only way to find out is to take a careful look at all the links in the chain, from Fx code to Windows DNS resolving, to Internet protocols. Possibilities include multiple tries and variations on DNS caching. Perhaps Fx is even caching negative lookup attempts. Maybe other browsers keep trying longer to get a DNS report. If so, how do they deal with Windows' caching of negative lookup attempts? Or do they? How could they possibly obtain an IP address if Windows refuses to try again, as MS documentation alleges?</del>

For the record, I just tried browsing with a modem. Fx successfully returned 17 of 17 Web pages, including the one mentioned in that bug report. The others were random pages from a Google search. They were loaded more or less simultaneously, in separate tabs.

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.0.7) Gecko/20060909 Firefox/1.5.0.7
Last edited by VanillaMozilla on September 18th, 2006, 10:00 am, edited 2 times in total.
User avatar
dickvl
Posts: 54145
Joined: July 18th, 2005, 3:25 am

Post by dickvl »

You have to remember that the google search uses link prefetching for the first result, so don't use the first but a random result.

You can disable the prefetching: network.prefetch-next
http://www.mozilla.org/projects/netlib/ ... g_FAQ.html
alta88
Posts: 1029
Joined: January 28th, 2006, 3:08 pm

Post by alta88 »

@trolly

very good. that's actually even less checking than i would have thought, not very exemplary design.

may i suggest the following:

internal retry based on pref/settable count.
after internal retries, display the *exact* error returned by the OS in the error dialog.
the error dialog should be in the form of Retry[countdown timer], Abort/Cancel; 10 second timer

intelligence could be built in based on the error, but this is enough for a first start. worthy of submitting a patch for sure.

millions may silently thank you ;)
trackback
Posts: 7
Joined: September 18th, 2005, 11:12 am

Post by trackback »

i also have ben getting alot of server can not be found and such, i was wondering what can i change in the "about:config" to disable this gray screen with the white box that tells me the server is not found and to try again.. i changed it before, i had to change something from true to false or visversa i forget..
so how do i get this error page not to show up, i'm pretty sure it didn't show up until i updated a while back.. i remember i changed it before but my computer crashed and my firefox went back to all default settings. thanks.
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

This one? network.dns.disableIPv6 -> true
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
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

alta88 wrote:@trolly

very good. that's actually even less checking than i would have thought, not very exemplary design.

may i suggest the following:

internal retry based on pref/settable count.
after internal retries, display the *exact* error returned by the OS in the error dialog.
the error dialog should be in the form of Retry[countdown timer], Abort/Cancel; 10 second timer

intelligence could be built in based on the error, but this is enough for a first start. worthy of submitting a patch for sure.

millions may silently thank you ;)
I do not think i can hack the code far enough to get the error number to the screen. But i have only scratched at the surface.
There is still this one about:config -> network.dns.disableIPv6 -> true
As far i have seen there is different code used if IPV6 is enabled.
fazza
Posts: 3
Joined: September 18th, 2006, 3:56 am

Post by fazza »

well i just reformatted again and put ff 1.0.4 on and all is well so i still say ff1.5.0.7 is at fault im sorry to be so blunt but it seems you guys are trying to make all the excuses in the book
VanillaMozilla
Posts: 13808
Joined: November 7th, 2005, 11:26 am

Post by VanillaMozilla »

Dickvl,
Prefetching is still fetching, and there was a case in which exactly the same test turned up over 40% failed attempts.
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

Don't complain if you catch some malware. 1.0.4 has several known security holes.
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.
VanillaMozilla
Posts: 13808
Joined: November 7th, 2005, 11:26 am

Post by VanillaMozilla »

fazza wrote:well i just reformatted again and put ff 1.0.4 on and all is well so i still say ff1.5.0.7 is at fault im sorry to be so blunt but it seems you guys are trying to make all the excuses in the book

Sorry, fazza, you haven't been reading, have you? Call it making excuses if you want, but Trolly just found a bug. Before you get excited, I'm not sure the bug applies to you. Read on.

As for your argument that the problem just started with this version, we've seen the same argument with version 1.5.0.6, then again with 1.5.0.5, 1.5.0.4, 1.5.0.3, ..., 1.0.6, 1.0.5, ..., 1.0, .... Each time someone argued that the previous version was fine, so they must have done something to this version. That doesn't add up. So Trolly is right. If you can fix the problem by going back to 1.0.4, it's almost certain you can fix the problem with the current version, and you won't be exposing yourself to security problems.

From your very brief description, it sounds like you have a firewall problem. Before you reply, read this article carefully.
User avatar
trolly
Moderator
Posts: 39851
Joined: August 22nd, 2005, 7:25 am

Post by trolly »

A bug? Not sure about that. I'll have to dig deeper to be sure and i can not see the problem at all. I'm just pointing out a potential problem.
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.
trackback
Posts: 7
Joined: September 18th, 2005, 11:12 am

Post by trackback »

trolly wrote:This one? network.dns.disableIPv6 -> true


thanks but i don't think thats it because the gray page still shows up, i'm talking about the error page specifily... not really the error...
i'm trying to go to this page www.musichq.nl but it says connection timed out and tells me to try again ect... remember a few updated ago when there was a error it never used to show this gray page with a white box.... thats what i'm trying to get rid of, this gray page and white box that tells me to try again...
by the way do you get an error too while trying to view that site?
Locked