Slashdot Mirror
Red Hat Linux 7 Infested With Bugs
TBHiX writes "Apparently, according to reports on bugzilla and on linuxnewbie.internet.com, Red Hat 7.0 is being described by some people as one of the buggiest distros they've seen in recent history." Red Hat's point-oh releases have been historically been pretty bad over the years, so I the only thing that surprises me is that people didn't realize it before they downloaded it. The point release has typically been fine, but the bugzilla report lists over a thousand bugs: 200 appearing this week.
Take this as a warning folks: didn't 4.0, 5.0 and 6.0 teach you anything? *grin* But a DB with 2500 bugs in it doesn't necessarily mean a buggy distribution either.
Bugs per line of code (LOC).
See Emphasizing Software Test Process Improvement... in it, they say:
So, If Windows 2000 was 30 million LOC, you can expect there to be, on average, between 120,000 and 180,000 bugs in the shipping code.
Let's be generous to MS, and say that they have an outstanding development process as decribed in the above paper. Because of this, they manage to reduce the number of bugs by a factor of 50%; so they're only shipping with 60,000 - 90,000 bugs.
Now, let's be even more generous, and assume that only 10% of the bugs actually present in a system is actually ever noticed and reported (BTW, a ridiculously low estimate, IMHO...) This means that W2K should have on the order of 6,000 to 9,000 reported bugs.
Now, contrast that with the latest Red Hat release; buggy as all git out, you know. 2000 reported bugs. You'd have to go back and compare LOC to get a comparable estimate, but I'm guessing that if you count all the various and sundry packages, Red Hat ships at least 30 million LOC in a distro... which would mean that their code, buggy as all sin and scorned by open source hackers everywhere, would contain about one-third of the bugs that W2K contains.
That's why you're buying Linux.
"Great men are not always wise: neither do the aged understand judgement." Job 32:9
<OSS_zealotry> .1? Fix 'em yourself.
Why wait for the
</OSS_zealotry>
Word to the wise: Don't upgrade if you want stability. However, for desktop users, you probably won't regret it.
(Yes, I know I should be using something other than RH on my desktop. It's just what I have used for quite a while, and I'm used to it.)
With as large of a user base as RedHat has (and increasing), there are bound to be hundreds of horrible stories. But for each horror story, how many more success stories go on that are untold?
Just because the number of bug reports goes up doesn't mean that quality of product has gone down. If a product gets 1 bug report and has 1 user, does that show its quality? No.
But with the RH7.0 release, how many hundreds of thousands tried the product? And we're looking at maybe 255 bugs? Probably less than 50 of which are genuine bugs.
It's easy to hear those who are having problems because they shout the loudest. It's the people who it works for that aren't heard. All the success stories are silent. I'm one of the success stories. I've installed RH7.0 on 4 servers now. (one 486, a P133, one P-3 850, and a big phat $15k dual zeon 550). My installs have been flawless; better than any distro previous (including 6.2). If everyone shouted their successes as loudly as the minority shouts their failures, then RedHat would be overwhelmed with positives (and deaf).
I give a standing ovation for the RH 7.0 release.
Daniel
The original "bug" is equivalent to someone replying to "I'm having (some problem) with Linux" with "I upgraded my Linux to Windows 2000 and the problem went away," here. I can't believe this was posted as a story; it's ridiculous.
Let me start with the "my views are mine, and do not represent my employer, blah, blah..."
Now, some very hard choices were made for this release, things like i18n support were desperately needed, and so we have a gcc snapshot as the complier. Somehow this makes us evil, whatever. People want 2.4.x asap, and It will likely land before the next release, so we linked against 2.4 headers, so that it can droped in later. Again, evil, I know, how could we even try to make upgrade easier?
Of course, an older gcc is provided as 'kgcc' so that you can compile your kernels.
Okay, what, exactly, do we ship from Red Hat? Why, we ship 2.5 GIGS worth of community developed software in Red Hat Linux 7. THe bugs we track, for the most part, for the OVERWHELMING part, are not in Red Hat written code, they are in community code. I say this with confidence, because most of the software is community code.
Let me ask you, how many times have you heard about a bug in 'screen'? The number is not small, and every one we know about goes into bugzilla, and gets closed when it is decided that either a) it is fixed, or b) we can't fix it. Now, screen is one of many hundreds of packages.
I am very upset with this style of journalism: "Red Hat 7 Infested With Bugs", honestly, is this a tabloid? Anyone who has ever used a bug tracking system KNOWS how this sort of thing goes, and most of those bugs are in everyone else's distros as well, the only differance is that people are shouting about our counting them. So ask yourself, what does it accomplish by posting a story with a title like this, knowing that CmdrTaco understands everything I've just said?
-- Crutcher --
#include <disclaimer.h>
-- Crutcher --
#include <disclaimer.h>
It's simple: Redhat (as several other distributions are now doing) is shipping a different compiler for compiling kernels than for general compilation.
You need to edit the kernel Makefile and add the line CC=kgcc (making sure that kgcc is installed on your system). kernel compilations will then proceed without problems.
Although this is surprising to many users, it's actually a good idea. The linux kernel stresses a compiler in ways that a regular program (even something like all of gnome) just can't do. Kernel compilation demands correctness of inlined functions, or preprocessor command parsing, of lock orders. Basically, the kernel uses C as a convenient macro language for assembly much of the time. The compiler's job is to faithfully translate.
Given all of that, the compiler that you want for general purpose compilation is not necessarily the compiler that compiles the kernel best. The kernel and compilers co-evolve. If you follow either the linux-kernel list or any of the gcc or libc lists, you will see that when bugs turn up with kernel compilation, it is as often the case that there is a libc or a gcc bug as a kernel bug. These bugs frequently only turn up in the context of a kernel compile (because who the hell else would do something like that?!?)
Finally, on the subject of the redhat release: RedHat did some dubious things (calling their gcc gcc 2.96 when 2.96 just doesn't exist, thereby forcing the gcc project to renumber their release to 2.97 just to catch any RedHat bugs, eg.). However, I'm steadily impressed by Redhat as a distro (flame me if you want, I don't care). Redhat is a commercial distribution but they release *all* the software they develop as gpl open source (and have set the tone among other distributors to do the same), and have struck a good balance between novelty and stability.
Anyway, that's how to solve your kernel compiling problem (among lots of other stuff).
You should know as well as me, that if you install in 1.5GB you install A LOT of unnecessary stuff.
Do a smaller install, or remove the unwanted stuff.
The reason it takes up more space is for instance that the SQL-part has been expanded with MySQL, as it has been made GPL. There are A LOT of new multimedia-applications etc.
If you installed without regard to this fact, you ended up with BOTH PostgreSQL AND MySQL, and this duplication in most other areas as well.
Please try to do a more minimal install before you scream "bloat".
Not to start a holy war.. But I really admire Debian for holding out and wanting everything perfect in a release, and I admire Redhat's approach of rapid frequency releasing.. But can't we get a middle ground? Either way, it's still fewer bugs that Microsoft Windows!
Story reads...
"Today, a newbie, in an experiment gone bad, allowed bugs to escape from bugzilla. They quickly escaped via the internet only to find a new host in RedHat."
Well, doesn't that suck. Sounds like something right out of one of the tabloids. A newbie pulls a bogus figure of his ass and gives it to RedHat.
Yesterday, I was comptemplating ugrading to 7.0 from 6.1 (which has run flawlessly for me). Then, I read some BS about 2500 bugs and instead decided to wait with my upgrade until 7.1. This morning, revised figures are posted by people that understand the process a wee bit more. Those new numbers come in around under 200 bugs.
200 bugs in a release as large as RH 7? Wow. If only most commercial software shipped with as few bugs. Take into consideration that most of these bugs are probably in community written code rather than RedHat written code. These are mostly bugs we've come to live with and are probably being patched as I write this.
What's a shame is that the original story will probably be picked up by MSNBC, CNN or some other news agency and the revised figures will not. Thus, people who may have upgraded or purchaced 7.0 may not do so now. Instead, they may continue to view Linux, in general, as buggy software that can not be trusted. Articles like this only perpetuate this myth.
I only hope that before posting potentially damaging material, that somebody actually check the figures before they are posted to the public. It strikes me as poor journalism, a disservice to the community and utter criminal negligience to do otherwise.
BTW...I'm going to upgrade to 7.0 today.
RD
www.linuxnebie.org and www.bugzilla.redhat.com
Computer 1: It's a Pentium III 550 HP NetServer.
SCSI 2 w/512 megs Ram.
I spent the whole of last night trying to get it installed over the network (didn't work). Would hang in the final process of the installation. I finally waited for the ISO to download and then tried to re-install the system w/the CD.
Everything seemed to work fine. I went through the installation process and when it came time to reboot the machine it FROZE SOLID on the 'Initializing Swap Space' message in init level 1.
I then went through and re-installed it again w/different partition settings , etc . 3 times total when I finally gave up.
I re-installed the machine with RedHat 6.2 and it worked fine. So much for my daring attempt at a x.0 release
...
I have a home machine that I use as a development server and it is a Athlon 750 w/512 megs of ram. Not a SCSI system but a UDMA/66 IDE System.
The machine was able to install correctly, but the interesting thing is durring times of VERY heavy stress to the system.. (Seti@Home or 3D Screen Saver) the machine hard locks. I have been running Mandrake 7.0 on this machine for 4 months and have never had a lockup.
I am guessing that there is a problem somewhere in the kernel that they provided or one of the libraries that they have compiled in that is causing some of these problems.
I have read as well in the kernel mailing list that there is a bit of hoop-da-la about the version of the compiler that RedHat uses.
Now, they are trying to get support in for the 2.4.x kernel as why I am trying to be brave in installing the system as a x.0 release. I believe as it is a bit of a moving target for redhat, there should be a bit of patience and support for them.
I do expect to see a lot of problems and issues as the 2.4 kernel roles out and everyone makes the necesary transitions to make everything run very smooth.
My opinion is simple. If you want rock solid, no bugs.. or at least worked out stable distro's.. don't go with a release that is not only a x.0 release, but a 1 week old release.
Use soemthing that is tried and tested.. such as Debian X (I am not a debian user, but they are a bit slower to release and tend to be more reliable (from what I hear))
The nice thing about RH 7.0 is you can help them out. You can install it yourself (assuming it boots.. (see case 1)) and re-install packages, recompile the kernel.. etc.
It is nice having all the latest and greatest libraries and such installed. It seems that the unfortunatley my RH 6.2 servers are going to be running RH 6.2 much longer than I anticipated based on my initial reaction though. (I will wait for the 7.1 version for any more non-private use..)
Be easy on RH as they do contribute a lot to the comunity. I don't believe that they should be lynched for taking chances on new technologies.
--------------------
I've been running Red Hat 7.0 for a little while. Things I've noticed:
/etc/rc.d/init.d, but have added /etc/init.d as a link for the Solaris users among us.
/etc/fstab: I dislike this. I used to go to /etc/fstab to find the device name of my various partitions. Now it just says "LABEL=/usr" instead of "/dev/hdb6".... This is consistent with the change to fstab where the label name is shown during fsck. I guess I understand the desire to display that information, but dammit, I want my fstab back.
.1 release, the fact that it's a .0 blows me away.
* I expected the first XFree86 4.0 release to be a little wonky, but it works well! Handled both my Permedia and Voodoo3 cards just fine with no user intervention (though, it's annoying that my monitor was not recognized, but it never was under the older X either).
* They've adopted a SysV/Solaris-like set of symlinks for the rc files. I like this. They still use the Red Hat style of
* Disk labels in
* Just the right balance of new vs old software. The kernel is 2.2 (wise, even though 2.4.0-test is pretty darn stable as far as I can tell); I hear gcc is a snapshot release, but I have compiled a whole hell of a lot with it so far; latest GNOME goodies are nice (not quite Helixcode nice)
* No problems yet. I've installed on two systems. One was having problems under both Windows and Linux, and the upgrade did not help, but did not hurt. The other was a test system, where I wanted to play with squid, and all worked just fine.
Things that scare me overall:
* Big distribution for Red Hat
* Semi-graphical LILO ala Corel
* xntp becomes ntp, which breaks a lot of scripts, and it's not on disk1
Looks good for a
Someone forgot to insert the leading http:// protocol identifiers for the links in the story, and MSIE 5.5 (here) is generating goofy URLs by inserting http://slashdot.org/ in front of them there links!
Proper URLs:
Sure they should be held accountable, if they've truly released a horribly buggy sytem. But it doesn't sound like they have. Alan Cox, in his diary says:
"The invisible and the non-existent look very much alike." -- Delos B. McKown
Oops. I meant to slashdot jwz :)
"Free your mind and your ass will follow"
Try 250 and climbing.
.before you post something that include a statistic
Seriously guys, at least check your numbers
-------------
Someday, you're going to die. Get over it.
A lot of work went into this release. I was on the beta team for this release, and there were a bunch of people working on a lot of different things to get this as ready as possible. I'm still running one beta version on a laptop and another on my desktop at home, and both are working just fine with everything I do with only two exceptions:
Are there bugs? Yep. Has there ever been a bug free OS release? Nope.
The Open Source movement - if you are to be believed, in its 'infancy' since the 60's - has plenty of both.
This is a detailed article on "the one bug".
- I don't care if they globalize against free speech. All my best free thoughts are done in my head.
I've gone from Red Hat 6.2 to 6.9.5 to 7.0. I submitted four or five bug reports in 6.9.5; one was a "already fixed in Rawhide" situation, two others got fixes a few days later, one has a fix coming when they put together a glibc-devel errata. XFree86 4.0.1 still consumes an ungodly amount of memory on my machine, but compared to the bugs in the first releases of Red Hat 5.0 and 6.0 (which I avoided until a few weeks of errata issues had gone by), I've been quite impressed. So far I've seen Red Hat 7.0 on 2 computers, and it's been good, 2 for 2. Your particular system may vary, of course. I'd advise waiting a month before upgrading all your corporate workstations, and waiting for 7.1 before touching any important servers... but if you're not in that kind of situation, come on in, the water's fine.
By contrast, I've seen significantly more problems with Mandrake 7.1, which was frighteningly down towards the Windows end of the quality-o-meter. That was a big let down, since Mandrake 7.0 had given our LUG such a smooth installfest last year. At least with Red Hat 5 and 6, the progression from "buggy" to "rock-solid" was steadily upward.
I don't know who pulled the number 2500 out of thin air, but a query of bugzilla as of 9:25pm on 10/02 shows "only" 149 bugs, and given the number of those that are NEW, there are probably less than 100 actual bugs. And of those, how many are RedHat's fault as opposed to buggy packages?
If someone pulled that number out of bugzilla, they must not have known how to use it. If not, then they just pulled the number out of thin air.
I'm not saying that 149 possible bugs is "good", but it is more in line with what you might expect a week after a major release.