WTF

Apr. 4th, 2006 09:12 pm
swestrup: (Default)
[personal profile] swestrup
It never fails. Just as I'm getting ready to call it quits for the day, Linux dies in some improbably manner and I am forced to continue working to resolve the issue. This time, I tried (for the Nth time) to boot a new kernel. It failed again, and so I rebooted back into the old one. Only this time it didn't come up.

It complained that there was no superblock found on /dev/md7. Now, thats the tmpfs partition, so its not like it really mattered, but I wanted to know what was wrong, so I spent some time sorting things out. Well, it turns out it was (for reasons unknown) building all the raid devices with the wrong stuff. It was trying to build /dev/md7 using the swap partitions, and was building /dev/md8 using the partitions for /dev/md7.  Say what?

Anyway, I managed to correct everything by booting off of a liveCD and manually building the raids then re-lilo-ing. I wouldn't have needed the liveCD except that mdadm seems to have no way to Unassemble a stopped raid.

Oh well, its now WAY past time to have supper. I was sort of planning to have finished digesting by now, and been off to bed.

Date: 2006-04-05 02:18 am (UTC)
From: [identity profile] thothmeister.livejournal.com
Your frequent headaches with Linux concern me, since I intend to make my PC dual-boot into SuSe Linux once it works again.

Date: 2006-04-05 01:37 pm (UTC)
From: [identity profile] sps.livejournal.com
I must admit I have no idea how he does it ;). My work machines generally break when I break them - and understand that my work does involve breaking them!

Date: 2006-04-05 02:32 pm (UTC)
From: [identity profile] sps.livejournal.com
It's true. The machines I work on, I basically only apply security patches or the odd missing library, because stability is so important. Maybe this means that there need to be two Linux boxen at your place, a production one and a development one! :-}

Date: 2006-04-05 10:03 pm (UTC)
From: [identity profile] pphaneuf.livejournal.com
When did you break something upgrading with YaST (YOU, actually)? The SuSE people are ultra-paranoid about updates, and basically only have updates for severe crashers or security fixes (and even then, generally backport the fix to the version of the package they shipped with the distro).

I remember talking to the maintainer of the Debian packages for the OCaml compiler (I think?), and he was explaining to me that since the OCaml people break the binary compatibility with every release (and they release fairly often, like every few months), combined with the OCaml system being separated in multiple packages (compiler, libraries, etc), dependencies between these packages and the Debian policies involving waiting a week or two for packages to be accepted, results in that every few months, OCaml is broken on Debian for two or three weeks at a time. If you do "apt-get dist-upgrade" during that time, or "apt-get install" OCaml, you'll break it.

Date: 2006-04-06 12:47 am (UTC)
From: [identity profile] pphaneuf.livejournal.com
I've been running YaST Online Update every night in a cron job for the last, hmm, four, five years or so, on a number of machines. I think I've had trouble once, where I had added some binary-only drivers to my kernel (with a tarball, not with a package!), and things got a bit confused. Was really my fault, I just re-installed the stock RPMs and re-updated and it set itself straight.

January 2017

S M T W T F S
1234567
891011121314
15161718192021
22232425262728
293031    

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Feb. 14th, 2026 11:53 pm
Powered by Dreamwidth Studios