Gentoo randomly crashes after installing for a few weeks

Message

Eidi · Post by **Eidi** » Sun Mar 26, 2006 7:08 pm

Argg... I don't know what I'm doing wrong, I've been installing Gentoo alot lately (I'm still a Windows user at heart, can't get over the "if it doesn't work, reinstall" crap

(But at least I no longer use Windows on my computer)) and probably a week or two after I've installed it, it starts to become really slow and actually hang. Dunno why. It's not like I'm using any exotic cflags or have alot of unstable packages. This is on an Athlon 64 X2 machine.

This is my make.conf:

Code: Select all

# These settings were set by the catalyst build script that automatically built$# Please consult /etc/make.conf.example for a more detailed example
CFLAGS="-march=k8 -O2 -pipe -fomit-frame-pointer"
CHOST="x86_64-pc-linux-gnu"
CXXFLAGS="${CFLAGS}"
MAKEOPTS="-j3"
USE="X kde qt alsa gnome gtk"
GENTOO_MIRRORS="ftp://ftp.ucsb.edu/pub/mirrors/linux/gentoo/ http://ftp.ucsb.ed$
SYNC="rsync://rsync.namerica.gentoo.org/gentoo-portage"

And my package.keywords:

Code: Select all

media-video/nvidia-kernel ~amd64
media-video/nvidia-glx ~amd64
net-im/gaim ~amd64
kde-base/kdebase-meta ~amd64
media-video/mplayer-bin ~amd64
media-libs/win32codecs ~amd64
app-emulation/emul-linux-x86-medialibs ~amd64
kde-base/kdelibs ~amd64
net-p2p/azureus ~amd64
dev-java/swt ~amd64

GAIM is the only one that's hard masked, but I don't think that would cause so much instability, would it?

Gaahh... I know I'm doing something wrong, but I'm not sure what.

Or maybe... You think it might be hardware?

I do things a bit weirdly, like I have /usr/portage/ on it's own partition, /home having it's own partition... But I don't think that would cause it.

Any advice on what NOT to do the next time I install? I borked it again and I'm just gonna reinstall.

P.S. I just use ext3 for all but /boot, which is ext2.

ectospasm · Post by **ectospasm** » Sun Mar 26, 2006 7:51 pm

Random crashes... would this happen after several days of uptime? My gut reaction is that this is a hardware problem. memtest86 used to come on the Live CDs (I'm not sure if the latest one has it; it would surprise me if it didn't). I would run that to make sure that you're not experiencing flaky memory problems.

Eidi · Post by **Eidi** » Sun Mar 26, 2006 7:59 pm

Nope, sometimes it happens shortly after a reboot (<1 hour), other times it takes a while. Been having problems with my nvidia kernel driver lately too, I think I can fix that, but with the other problems I've been having, and not liking gnome, I'll just do a reinstall.

I'll try memtest86 tonight, it takes a while to run, right?

Edit: I had a crash earlier, a totally hard crash, couldn't even Crtl + Alt + F1 to a console to kill something. It happened as I was just browsing the web, downloading something with Firefox (And trying to download something else, but it hung partway through, so I then used wget) and it crashed shortly thereafter. I really think I borked my install again.

ectospasm · Post by **ectospasm** » Sun Mar 26, 2006 8:07 pm

Yeah, memtest86 takes a while to run, you should leave it on all night to have several passes. I'd suggesting turning the scroll lock on; of course this will pause the test if the number of errors fills the screen, but at least you'd see the first errors that occurred.

razze · Post by **razze** » Mon Mar 27, 2006 5:01 am

I have a similar processor and use the same CFLAGS, so that does not seem to be the problem.

A small tip: you might want to set MAKEOPTS to "-j5" (this settings determines how many simultaneous compiling jobs the compiler does, and a good rule of thumb is # of processor x 2 + 1) to speed up compiling a bit. The difference will not be huge, but at least it won't slow you down either.

Razze

whig · Post by **whig** » Mon Mar 27, 2006 5:09 am

Check those logs. After a freeze boot up a knoppix disc, go to /var/log/ of your system and type

Code: Select all

ls -lrt

the most recent will be at the bottom. It will also pay to install magic sysrq keys if you haven't already,
http://aplawrence.com/Words2005/2005_04_13.html

ectospasm · Post by **ectospasm** » Mon Mar 27, 2006 6:55 am

razze wrote:A small tip: you might want to set MAKEOPTS to "-j5" (this settings determines how many simultaneous compiling jobs the compiler does, and a good rule of thumb is # of processor x 2 + 1) to speed up compiling a bit.

Is that really considered a good rule of thumb? It used to be #cpus + 1, so a 2-way SMP would be -j3 (that's what I have). I could be wrong. I also come from the school of thought that swap should be twice your RAM, and I hear that's not considered to be the accepted rule of thumb anymore, so who knows? OK, enough of this OT discussion.

mudrii · Post by **mudrii** » Tue Mar 28, 2006 1:13 pm

can you post emerge --info

Eidi · Post by **Eidi** » Thu Mar 30, 2006 4:21 am

I think it was a hardware issue. Changed a few things (Such as revolting my processor up higher, forgot about that. And resetting my RAM to the auto value which is DDR333, but the RAM is rated at DDR400, so I dunno) and it works fine now.

Mandr4ke · Post by **Mandr4ke** » Fri Mar 31, 2006 3:12 am

i've been running amd64 with my new hardware for over 4 months now without a single crash or freeze.

Although my old computer useta be plagued. High probably it's usually always hardware!

1.st Check if your system is over heating. jump in the bios and if your temp hits 60c then your in trouble.

if it's not over heating, then high probabilites are power supply or video card, memory. I've had a few bad viedo cards before that would make even windows reboot once an hour driving me nuts for a very long time, if you have a spare vid card / memory, power supply swap it out till you track it down.

BTW: you might want to double check your BIOS settings and make sure they are not goofed, they can be tricky on new systems now a days.

Eidi · Post by **Eidi** » Sat Apr 01, 2006 7:48 am

Mandr4ke wrote:i've been running amd64 with my new hardware for over 4 months now without a single crash or freeze.

Although my old computer useta be plagued. High probably it's usually always hardware!

1.st Check if your system is over heating. jump in the bios and if your temp hits 60c then your in trouble.

if it's not over heating, then high probabilites are power supply or video card, memory. I've had a few bad viedo cards before that would make even windows reboot once an hour driving me nuts for a very long time, if you have a spare vid card / memory, power supply swap it out till you track it down.

BTW: you might want to double check your BIOS settings and make sure they are not goofed, they can be tricky on new systems now a days.

I need to get a burner to reburn a LiveCD to test, but I'm thinking I have bad memory because it's auto detected at DDR333, but is said to be rated at DDR400, so I try to manually set it to DDR400. I reverted it and it's working fine.

Maybe I got ripped off on my memory. Bought it from NewEgg too, damn...

Also, I live in Arizona, it does get pretty hot here, but I set my BIOS to auto turn off my processor when it hits 60 celcius.

ectospasm · Post by **ectospasm** » Sat Apr 01, 2006 8:08 am

Eidi wrote: I need to get a burner to reburn a LiveCD to test, but I'm thinking I have bad memory because it's auto detected at DDR333, but is said to be rated at DDR400, so I try to manually set it to DDR400. I reverted it and it's working fine.

Maybe I got ripped off on my memory. Bought it from NewEgg too, damn...

Also, I live in Arizona, it does get pretty hot here, but I set my BIOS to auto turn off my processor when it hits 60 celcius.

If you ever think you're having memory problems, it would be in your best interest to run memtest86 (it's on the LiveCD, or at least it used to be). It works with amd64, and if you let it run for a few cycles, it will tell you if you're having memory problems. Does better than a lot of the commercial offerings, or so I hear. I'd suggest running it at the speed you wanted to (400), and see how many memory errors you get. Then run it at 333 and make sure that there are no errors there.

I could imagine that you accidentally bought 333, Newegg sometimes goofs on their product labelling; I almost bought a PCI-E nVidia card from them when what I needed was AGP, because they had mislabeled one of the titles. If you still have the receipts for it, send it back if it clearly says it should be 400. Of course, memory producers are shady and underhanded, it may not be Newegg's fault.