FS#4821 - pacman doesn't install kernel-headers package in testing properly on -Syu
Attached to Project:
Pacman
Opened by kongokris 2 (nut543) - Wednesday, 14 June 2006, 17:59 GMT
Last edited by Jan de Groot (JGC) - Thursday, 22 June 2006, 21:23 GMT
Opened by kongokris 2 (nut543) - Wednesday, 14 June 2006, 17:59 GMT
Last edited by Jan de Groot (JGC) - Thursday, 22 June 2006, 21:23 GMT
|
Details
Today i decided i wanted to try /testing.
So i enabled testing and moved it to the top of pacman.conf then i did sudo pacman -Syu and then since i got an error that pacman couldn't get correct filesize for mkinitcpio but still continued it's operation(i consider this a second bug btw) but ignoring mkinitcpio, I control C'd the operation and then did a new pacman -Syu This time mkinitcpio downloaded correctly and the few packages that wasn't already downloaded from the previous -Syu downloaded fine, and the whole installation procedure finished without a hitch. fast-forward a reeboot and some hours. kfusd gives me a bunch of errors! after a hint by some guy in #bash i reinstall kernel-headers and now 99% of the errors disappears! info: kris|/var/abs/local/fusd-kor$ uname -a Linux garbagepailkid 2.6.17-rc6-ARCH #1 SMP PREEMPT Tue Jun 6 10:18:20 CEST 2006 i686 GNU/Linux kris|/var/abs/local/fusd-kor$ gcc -v Using built-in specs. Target: i686-pc-linux-gnu Configured with: ../gcc-4.1.1/configure --prefix=/usr --enable-shared --enable-languages=c,c++,objc --enable-threads=posix --enable-__cxa_atexit --disable-multilib --libdir=/usr/lib --enable-clocale=gnu Thread model: posix gcc version 4.1.1 kris|/var/abs/local/fusd-kor$ pacman -Qs glibc kernel-headers local/glibc 2.4-2 GNU C Library local/nss-mdns 0.8-1 glibc plugin providing host name resolution via mDNS local/kernel-headers 2.6.12.0-1 Kernel headers sanitized for use in userspace kris|/var/abs/local/fusd-kor$ pacman -Qs xfs local/acl 2.2.34-1 Library for filesystem ACL support local/attr 2.4.28-1 Extended attribute support library for ACL support local/dmapi 2.2.3-1 Data migration API local/xfsdump 2.2.33-1 Additional XFS filesystem utilities local/xfsprogs 2.7.11-1 XFS filesystem utilities local/xorg-xfs 1.0.2-1 X.Org X11 xfs font server everything is -Syu'd to the latest version. This to prove i installed it twice: kris|/var/abs/local/fusd-kor$ grep kernel-headers /var/log/pacman.log [06/14/06 14:38] installed kernel-headers (2.6.12.0-1) [06/14/06 19:23] upgraded kernel-headers (2.6.12.0-1 -> 2.6.12.0-1) probably bug #3 i get :: extra is up to date error: anonymous login failed now when I do -Syu. but this and other i will create secondary bug report if i can be bothered. |
This task depends upon
Closed by Jan de Groot (JGC)
Thursday, 03 August 2006, 21:25 GMT
Reason for closing: Duplicate
Additional comments about closing: See 3999
Thursday, 03 August 2006, 21:25 GMT
Reason for closing: Duplicate
Additional comments about closing: See 3999
For the kfusd thing: what is it? AFAIK it isn't in any of our repositories. What errors did you get, etc.
Then the kernel headers: the only case where problems with these could happen is with a pacman version that is dated 2 or 3 releases ago, or with a forced glibc downgrade back to 2.3.6 after trying 2.4 from testing.
I've redownloaded pacman as well now incase there was a mis-install of some kind in the first place.
i'm using Server = ftp://ftp.archlinux.org/extra/os/i686.
when files move between packages and pacman updates in incorrect order?
For further information the issue appeared after moving testing repo to the top of pacman.conf and uncommneting it. then doing a pacman -Syu which also included the "error: anonymous login" error.
It then finished downloading almost all packages but I ctrl+c'ed it because after being unable to "get filesize" for mkinitcpio package and seeing like pacman was then just going to ignore it: I sudo pacman -Syu'ed and now mkinitcpi downloaded and install of all packages went "seemingly" without a hitch.
i had about 2.5 Gb(this number i can't remember for sure, might have been more) avalable on a 21 Gb XFS partition available before starting the 450ish Mb total of pacman files.
kris|~$ cat /etc/mtab
/dev/hda6 / xfs rw 0 0
none /proc proc rw 0 0
none /sys sysfs rw 0 0
none /dev/pts devpts rw 0 0
none /dev/shm tmpfs rw 0 0
none /proc/sys/fs/binfmt_misc binfmt_misc rw 0 0
was the systems open at the time i assume.
I've included
egrep '14\/06' /var/log/pacman.log > 14-6-06.pacman.log
egrep '15\/06' /var/log/pacman.log > 14-6-06.pacman.log
as i continued into the night.
I'm thinking maybe this is a HD error/software error which triggers this because when the partition hasn't got so much room left the write head of the HD has to move much faster and if it is a lots of files to install like this then the write head might errounously "skip" a few writes?!
I've had no problems with this HD though.
In addition to that all us XFS users will get a speedier pacman not to mention all the laptop users with slow/5200 RPM HD's. Don't forget us!
I have seen the exact same problem with my XFS machine.
It just happened today after pacman -Syu'ing (only coreutils and lua got updated.)
(Big update forced me to reinstall all packages... Better sure than sorry.)
This ment for me that coreutils didn't get installed correctly and my X stopped responding. Kernel compile (makefiles) stopped working too... all of a sudden mv and rm were missing.
[438][PETER:#] tail -4 /var/log/pacman.log [/home/benny]
[06/25/06 13:32] starting full system upgrade
[06/25/06 13:41] upgraded coreutils (5.96-1 -> 5.96-2)
[06/25/06 13:41] upgraded lua (5.1-1 -> 5.1.1-1)
[06/25/06 13:41] upgraded coreutils (5.96-2 -> 5.96-2)
This did not happen on my jfs machine.
(I tried to see what happens when I pacman -U <path_to_coreutils>
again it killed half my machine... su stopped working for a moment... command not found, and even startx was not found... (Which doesn't even is in the coreutils package)
Maybe pacman triggers a xfs bug?