fink killed my comp?
 

Hello,

Well I'm not sure it was fink, but here's the thing: The computer has been running wonderfully for months, not one crash and all was great. I had the usual set of applications open.

Today I come home from class and I see the strangest dialogue box. It looked like the dialogue box when you install the computer, everything was greyed out and I had this nice window, in six languages, that basically told me "you need to restart your computer, press and hold the power button, yadda yadda".

Now, the only thing I was doing differently today is that since yesterday I was updating all my fink packages. I did a fink selfupdate-cvs and fink update-all. I'm running jaguar and already followed the instructions for updating to jaguar long ago (the one's sao posted, or linked to).

I checked my logs and I don't see anything, but I'm not exactly sure what to look for. Anyway, when I rebooted the first app that launched (mail) quit unexpectedly; also none of my other partitions/drives mounted. So I restarted and ran DiskWarrior.

After that I restarted again, fsckd the drive and continued to boot. So far I haven't had any other problems and it seems to be back to normal.

a. What could of caused this? was it really fink?

b. Is there any way of checking if fink finished compiling? What would be a good way of deinstalling fink and reinstalling it from scratch just in case.

c. Any other preventive troubleshooting things you can recommend?



thanks,
V

vonleigh,

There were other reports of the same "white box with the message saying that you needed to shut off the computer. The message was in English, German, French and Japanese" by osxpez and mervTormel. Check this thread:

http://forums.macosxhints.com/showth...ghlight=Jaguar

There were no explanations of why it happen or solutions, but, I doubt very much that is caused by Fink.


Cheers...

Re: fink killed my comp?
 

This is the new face of a kernel crash.

I have had several fink-associated kernel crashes which I believe is due actually to a gcc problem because it occurred (reproducibly) when gcc was called to compile something in qt. My problems resolved when some changes were made to several packages within the /unstable tree.

Are you by chance updating a version of kde? Were all of your packages installed under Jaguar or did your /sw installation predate Jaguar? If the latter is true, I would definitely reinstall fink and rebuild all of your packages.

I had a number of problems with a recent fink installation where kernel crashes happened and trashed my boot hard disk twice. One thing you need to do is at least run fsck from single user mode, i.e., reboot holding down apple-s, type /sbin/fsck -y. If fsck runs and does not report that it cannot repair something, you should run /sbin/fsck -y AGAIN and see if it reports the same errors. If everything is OK, you should see no errors on the second running of fsck. If fsck gives you troubles, I would recommend backing up everything immediately and running a disk utility. My particular problems were not repairable by Norton Utilites OR Disk Warrior, and I ultimately reformatted my drive. Because of all these difficulties, I subsequently made a new partition for /sw and mount this at boot. No problems yet....

Hello,

Thanks sao and braindead for your answers.

Yep, I guess it was definitely while compiling; didn't know that was the new face of a kernel panic, I guess it's less intimidating for newbs.

My /sw definitely predated jaguar, but I had already updated it once before (after I followed the update instructions to jaguar).

I think what I'm going to do is deinstall fink and reinstall it over, see if I don't have any problems this time. What would be the best way of doing this? Simply rm'ing the whole /sw directory or?



thanks,
V

vonleigh,

Please wait a while before deleting and reinstalling. I have posted about the problem at the fink-users list, and we might get some feedback soon.

So, let's wait some...

As I'm writing these words, came the first answer (from Ben Hines):

<<Just a kernel panic. I've been getting a lot of them recently, i believe caused by my EyeTV.

Fink doesn't cause kernel panics, but it can sometimes get Norton Antivirus to do so. Not fink's fault, Norton's.>>


Cheers...

BraindeadMac,

As I recalled, we already talked about your problems, you had a very messed up installation of Fink mixing and installing binaries from 10.1 in 10.2. Fink didn't really messed up your installation, but you did.

http://forums.macosxhints.com/showth...&threadid=6679


Cheers...

vonleigh,

As I see it, maybe it's not necessary for you to reinstall Fink at all.

I also have like you an installation of Fink (/sw) that predated Jaguar (actually from December 2001) and upgraded to Jaguar same as you did. And my installation works quite well.

Let's check the basic from yours, same as we did in another thread with Bluehz.

'fink -V'

'fink list --section=x11-system'


Cheers...

Well, we did talk a lot about that and I realize it was a complicated installation of fink in one instance. But I can tell you that I had kernel panics building kde with no 10.1 binaries installed.

And no, I disagree with your assessment that I messed up my hard disk. That was caused by a kernel panic, which was caused by gcc. Not by fink, we agree on that.

And no I don't have Norton Antivirus running (or Virex, for that matter).

BraindeadMac,

Yes, you are right. Well, it was a complicated installation.

But hey...nothing wrong with that. :)


Cheers...

Sao,

I don't have an EyeTV or anything of Norton installed (I strongly encourage people _not_ to install norton and to run it only from the cd).

I'm pretty sure what crashed it was the act of compiling, because the computer has been running for months and no other changes (besides compiling) have been done, and no other different programs were running aside from the usual. Also, in that thread you referenced me to the poster also said he was compiling.

I was considering reinstalling because, even though I'm not sure why it crashed, i'm sure it crashed before it finished compiling, therefore I'm assuming that somewhere some program is a bit munged.

Here's the result of:


I really don't mind reinstalling it from scratch, what I'd like to know is the more correct way of doing this. I already did a fink list and wrote down all the packages I installed.



thanks,
V

vonleigh,

Your installation seems ok to me.

Well, If I decided to remove Fink and reinstall, I would do the following:

First, I would run this:

sudo dpkg --force-depends --remove passwd g77

To remove the passwd package and g77.

Also, there are some packages that install files outside of /sw that you may need to remove with 'fink remove' before deleting everything. The ones that I know of that could cause problems are:

The postfix(-release) package moves the default sendmail
The daemonic package puts items in a StartupItems folder

Then to remove Fink:

sudo rm -rf /sw

and, if you have XFree86 installed from fink:

sudo rm -rf /etc/X11 /usr/X11R6 /Applications/XDarwin.app

Then I would follow the instructions here to install Fink from scratch in 10.2:

http://fink.sourceforge.net/news/jag-bootstrap.php


Cheers...

Thanks sao,

I'll get to it right now, I'll inform if anything weird happens.

BTW, ever since that crash my computer has been wonky. Today I had to restart because everything was giving me errors, basically anything net-related gave me permission errors. Tried opening the terminal for some troubleshooting and it asked my for my password? (no login message or anything, just "password:").

I restarted and I found in my logs messages that indicated that netinfo couldn't contact localhost. I hope deinstalling fink helps, but if not i'm not sure if i should clean install jaguar and start over, ::sigh::



V

BACKUP YOUR DATA NOW! You probably have a major disk error secondary to the kernel panic.

FINK IS NOT YOUR PROBLEM. Sorry, had to shout that. Remember fink installed software does absolutely nothing unless you tell it to. If you want to make fink "disappear" then you can just take out the "source /sw/bin/init.sh" instructions from .tcshrc or wherever you've put it (and of course, exit the current terminal shell instance). If you have specified the /sw/bin and /sw/sbin as part of your path anywhere else, you'll need to fix that too.

The kernel panic has trashed some disk data, I would suspect the netinfo database with that log error message.

YOU MUST RUN FSCK AT LEAST TWICE from single user mode. Sorry, had to shout twice. but all these people keep removing fink when it is not necessary. I explained this previously in my post. And no, I'm not a nut or idiot. If fsck does not fix the disk errors you must run DiskWarrior. Norton Utilities will choke on the disk problems most likely. DiskWarrior may choke on them too (did in my case) in which case you will have to repartition your drive (clean install of jag will not fix disk errors without a reinitialization.

If fsck or Norton or Diskwarrior fix your problem, then you still need to repair the netinfo database. Do a search at apple knowledgebase for "netinfo database" and you will find the instructions to do this in jaguar.

Braindead,

Thanks for your concern, but if you read up you'll see that I already ran DiskWarrior and repaired my main drive

Additionally I've already run fsck on the main drive in single user mode and disk utility on the other drives/partitions. In all cases they reported no problems.

After doing all this I had the problem described in my last message with netinfo.


V

KDE installs?
 

I have seen this happen during KDE (3+) installs during the compilation process. I have since totally given up on Fink, though on the linux side, I am a Debian fan. I no longer use a "/sw" directory.

I have since went to BSD Ports. I no longer use Fink for Darwin. I usually install and compile source into the BSD subsystem myself [usr/local for most] and have never had a OS version conflict (10.2 now). I just make sure that the new gcc can handle things before installing certain binaries. I may reconsider Fink at a later date after they iron out the conflicts and the binary is ready...., maybe in 2010.

It's very easy to just talk and bash Fink. Can you show some proof? Where are your error messages of your KDE compilation.

Please, you are welcome to post them here. I absolutely don't agree with what you say. Fink doesn't create kernel panics.
Sarcastic, and ungrateful, isn't it... The Fink developers are working very hard, a great bunch of people working for free in their spare time, they already gave us lots of applications we can install from source right now. I personally, have installed more than 700 packages with Fink, as many others did, and besides, the Fink developers will come up with the binary release very soon. Instead of speaking like you did, if you would know how, then the best way would be to give them a hand, in order that we all can benefit from it.

I don't like cheap talk, specially when Fink has 1238 packages in 22 sections listed in the Fink database as of November 09, that are ready to install.


vonleigh,

I agree very much with what BraindeadMac said that Fink is not your problem. And with his advice, specially when he had a similar experience. But I also understand the goodness of making a clean install.


Cheers...

That was a QT bash, that is, if you know how to read between the lines.

I know that the date 2010 was sarcastic. Maybe I should have said 2005.

I wouldn't know if it is easy or not to bash Fink. I really haven't tried. I too used Fink since the beginning, but now that I know how to "roll' my own, I no longer need it.

However, it would be quite easy to bash you. Consider yourself bashed. In fact, I feel more inclined to zsh you, so then %%$$##***,,, consider yourself zshed too. Be careful or next time I'll Korn you.

Glanz; if you don't like fink, we won't make you use it.

But if you expect to get any help from people here, try to keep it confined to the problem and not the person, OK?

Craig,

I thought he was just trying to be humorous, I hope sao doesn't feel ofended by it; as he helps alot in the forums and is a great fink champion.

So far I haven't had any other trouble, but I am on my toes and keeping my backups more regularly.

The odd thing is that before this fink update my computer was running beautifully for months. It runs apache, sendmail and all kinds of goodness and stays on 24/7. Just while compiling it crash and died and I experienced all I wrote about in this thread; that's why I'm blaming the compiling process.

Please don't take this as anything against the fink developers, I am very grateful for their great work and continue to use fink, although I have a lot less packages installed.

If I get any other weird errors from here to the next few weeks I'll reinstall jaguar just to be safe. The reason is that in december I'm going to leave this server unattended for about a month, and it has to stay up and working.

Up until last week I was sure it could, as it had been up for over 3 months with not one problem.



V

Evidently, you can't read either! I never said I didn't like Fink. The other idiot did.

# If there's an INTELLIGENT person out there, prod him a bit.

if tty -s; then
echo Enter text end with \^D
fi

# No answer!

Great, I stick up for you and you turn out to be insulting anyway. What other idiot? The only people that posted in the thread are me, sao, and braindead, but he didn't mention anything about fink.

These forums I like because most people are nice, it's not a flame fest like others; I sincerely hope you reconsider your attitude.


V

Glanz; allow me to rephrase.

You call someone here an idiot again and you may become our first person to be banned. Am I clear about that? Knock it off or take it someplace else.

Hello,

I've been getting strange errors ever since the crash, and I wonder if anyone knows what they may be related to:

1. I get random "permission denied" while accessing webpages and even checking my mail.

2. When starting up sendmail I get the following error:


Nov 11 14:04:34 alpargata sendmail[391]: restarting /usr/sbin/sendmail due to signal
Nov 11 14:04:35 alpargata sendmail[1898]: starting daemon (8.12.2): SMTP+queueing@01:00:00
Nov 11 14:04:35 alpargata sendmail[1899]: gA80LBMP006070: SYSERR(root): readqf: ./qfgA80LBMP006070: line 1: bad line "0"


I've never seen this error before, and googling it didn't yield anything pertinent. I appreciate any info on how to fix this so that I don't have to reinstall jaguar.


thanks,
V

re: #2

http://lovecraft.die.udec.cl/orielly...il/ch23_03.htm

see entry: 23.3.4 Unknown Control Character in qf File

---

you may need to delete bogus items from your sendmail queues in /var/spool/

show us:

% mailq -v

Hello,

Uhm... seems that way:


and everything looks kosher in /var/spool, but:

Are those the offending files? If so, should I just delete them?

Also, I just discovered new errors, now popper is giving me EOF errors:



thanks,
V

yep. yep.

mailq says no control file, but the qf is the control file and the df is the data file. hmm... well, look at the qf file and see what's in it. be curious.

dunno about popper... might be related; anything in /var/spool/clientqueue ?

lol, it was spam... how appropriate. Clientmqueue is emtpy (besides a .pid file).

Let's see if this does the trick for popper.



thanks,
V

vonleigh,

Just to let you know, if you want still to reinstall Fink, that I read a post by Max Horn at the 'fink-dev' list saying that they will probably release Fink 0.5.0 in between 1~4 weeks time.


Cheers...

Hi sao,

Thanks for the info, but I already did bootstrap fink, can't live without ncftp, fortune and all the rest of the goodies. I haven't reinstalled x though, as I really wasn't using it that much.

I guess I'll get into x after I make my chop box (see coat room for info).



V