serious stability problems! please help!

ExecutorElassus

okay, so my computer just got moved on a long road trip. after setting up in its new home, i put in a wireless card. that still isn't working, but that's another story.

lately, as in the past day or so, i've been having MAJOR stability problems: opening konqueror to browse my hard drives runs the CPU up to 100%, and hard-locks the system. commands take much longer to execute. when i ran an 'emerge --sync' today, i started getting really weird errors about I/O errors, and about how the file system was 'read-only.' now, on booting, certain programs (always the same ones: cupsd, ntpd) fail to initiate, returning a "bus error." vixie-cron returns an error about a 'read-only file system,' too. to top it all off, X won't start anymore, at least not in a stable way.

this is all really freaky, yes? i have one hard drive that's 100% full, which is a recent thing. other than that and the wireless card, i can't think of what could be causing all the stability issues. the computer now takes about 10 minutes to boot, where i used to be able to boot in under 2. something seems to be seriously bogging down the computer, even though top shows a cpu usage of under 5%.

i hope someone can help me here. i'm really worried.

thanks,

EE

Dammital · Apprentice Joined: 05 Nov 2004 Posts: 189

At the expense of belaboring the obvious, you have been in the can installing a wireless board. So open it back up again and double check the HD cables, and make sure they're all seated properly.

You have a disk that is completely full? You want to do something about that. Again pretty obvious, sorry.

You may get some mileage out of running a manual fsck against your offending filesystem. (May not!)

Otherwise I've got nothin'.

ExecutorElassus · Posted: Mon Oct 16, 2006 12:50 am Post subject:

so, i did

Dammital · Apprentice Joined: 05 Nov 2004 Posts: 189

ExecutorElassus · Posted: Mon Oct 16, 2006 3:13 pm Post subject:

so, the first run through with e2fsck went overnight, and when i came back, the monitor was getting no signal. so i had to reboot. doing it again now, and nothing seems to be happening.

this, my /usr partition, seems completely b0rked. so, the next question is: what's the best way to migrate each partition of the drive to a new hard drive, and how can i replace /usr? i think i might be able to get files off the drive, so what's the process? what's essential, and what can i simply replace with an 'emerge world'?

please help, folks, my computer's been stuck in maintenance mode for two days now.

thanks,

EE

Dammital · Apprentice Joined: 05 Nov 2004 Posts: 189

chrbecke · Posted: Mon Oct 16, 2006 4:12 pm Post subject:

1) Make sure you've got backups of your important data.
2) Make backups of /etc and /var/lib/portage/world, extract a stage3 over your system, restore /etc and /var/lib/portage/world from backups and do an "emerge -e world". Luckily I haven't had to try this, yet, but it should work.

Good luck!

ExecutorElassus · Posted: Mon Oct 16, 2006 10:59 pm Post subject:

okay, here's some more info. i want to figure out what's going wrong, because my computer stopped booting (not even giving me the BIOS startup page). i swapped the memory, and now it boots, but it still is REALLY slow. running top, i see that any time i try to run a process (say, a command), the '%wa' marker goes up to 100%, and sits there for a few minutes. this - according to some guides i read - refers to 'time spent waiting for i/o.

now, why is it doing that? is it really a busted hard drive? or something else?

can someone give me some advice? i want to fix the problem (or at least know what it is) before i waste a bunch of money on hardware.

thanks,

EE

gentoo_lan · Posted: Mon Oct 16, 2006 11:01 pm Post subject:

Sounds like a busted harddrive to me. I would suggest backing up important data and getting a new harddrive.

ExecutorElassus · Posted: Mon Oct 16, 2006 11:10 pm Post subject:

got the new hard drive already. 250gb to replace 60; should be nice. so, like the other guy says: back up /etc/* and world, and then do a stage3, right? lemme ask a couple questions:

can i just use tar to copy those over? i can't remember what's on all the partitions of that drive (i know a couple: /usr, and a subdirectory of /home). can i just tar them into their new partitions?

how do i just do a stage3? boot from a liveCD, then copy the stage3 tarball, then what? uncompress it onto the new /usr partition?

sorry, but i haven't done this in a few years. i'm rusty.

oh! and even better! it won't boot with the RAM i have, so i have to go buy new RAM, too. grr...

anyway, thanks for the help

EE

ExecutorElassus · Posted: Thu Oct 19, 2006 7:58 pm Post subject:

okay, so i answered some of my own questions about tar-ing, and stuff. sorry for the unnecessaries. now, though, i have a new problem. whenever i try to emerge something, i get this:

ExecutorElassus · Posted: Fri Oct 20, 2006 8:37 pm Post subject:

hey, chrbecke, i have a question for you (or for any others who might know).

so, i've backed up my data. now, the only remaining partition on the drive is the one that holds /usr. since i don't want to use that partition, and i want my new /usr partition to be on a new drive, how should i unpack the tarball? would it be possible to unmount the old /usr partition, mount the new (empty) one, and untar the tarball into it? or do i need to move the tar and bunzip2 executable somewhere safe, like /sbin?

sorry for trying to do this the hard way, but i really want to avoid starting completely from scratch.

thanks,

EE

ExecutorElassus · Posted: Tue Oct 24, 2006 12:36 am Post subject:

okay, i have another problem. so, i have /dev/hda, which is the bad drive. GRUB is loaded onto this drive's MBR, and i want to install it instead onto my new drive. what i'm doing right now is booting off of /dev/hda, and when i get a prompt, installing grub on the new drive, which is /dev/hdc (using 'grub-install /dev/hdc'). then, i shut down, put hdc into hda's slot, and try to reboot. when i do that, all i get is

ExecutorElassus · Posted: Fri Oct 27, 2006 12:20 am Post subject:

all right, so i'm giving up trying to get GRUB to work for now. here's another problem:

when i try to emerge world, the first thing it does after installing the first program, is to delete all the obsolete ones on my system (i unpacked a stage3 tarball, and restored my /etc directory). once it does that, though, emerge no longer works: i get an "error with sandbox binary" message, and portage quits out.

is there a way to prevent portage from deleting older versions of programs unless they've been recently updated? as in, only delete the old version of the program currently being emerged?

thanks,

EE