DebianThis forum is for the discussion of Debian Linux.
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
I'm writing from Windows from another machine while a dear old Knoppix is trying to repair my Debian testing system...
I'm gonna tell you what's happened.
Yesterday evening a ran an "apt-get update" and saw some new packages (expecially the ones concerning to a final full upgrade of Gnome to the 2.12 version), so I decided to upgrade only some of them (yeah, because the full upgrade would be about 280MB...).
By upgrading "gnome-volume-manager", I was noticed that dbus and hal would need to be upgraded too. That's ok. Just after the debs got downloaded, APT started to install/configure the new packages by stopping some services and set the new changes. I noticed that dbus produced an error message regarding the parameter "--drop-privileges", so I thought the new installed version wasn't supporting that parameter anymore (quite odd, but quite possible at the same time).
Unfortunately, I had to go away a little time and when I came back I forgot to investigate about that message, and I turned off the PC.
This morning, while the kernel was booting, I read the Evil(TM): the dbus server (loaded by init scripts) produced a segfault. Just after, the console get f**ked and all the characters became unreadable (the same thing that happens when you cat a binary file and you have to reset the console), while the boot process was trying to finish. Magically, this mess ended up into the login form (yeah, even if it was something like "#[@§_:"), but when I tried to login, I was told that I couldn't login because the boot process was still in progress!!! So, I rebooted the machine, but I got a new result: "Kernel panic! Unable to mount root fs blablabla...".
Ok, let's summon my old Knoppix and try to fix the trouble.
Now, with a general check, I realized that:
1) my "/root", "/usr" and "/var" partitions (each one with ReiserFS) are totally unrecognizable and broken (the "/home" partition is safe because boot process failed, for some unknown reason, to mount it)
2) there's no way for me to chroot into my Debian system
3) cfdisk and fdisk say that the partition table is ok, but the superblocks of each partition are busted
4) there's no way for "reiserfsck" to repair the damage (I tried all the options, now I'm trying "--rebuild-tree" and and it took more than one hour to get to 60% of a 70GB partition)
Welcome to Hell. Now I'm waiting for reiserfsck to finish, hoping to get some good results, otherwise I'll have to format those three partition and loose tons of data.
So, be careful with the updates of Debian testing, expecially with dbus and hal.
If you have any suggestions to help me, they are absolutely welcome.
It could be intersting to have a look at /var/log/dpkg.log to see what was installed.
Because I don't see any Critical Bugs on packages.debian.org and because second I just updated hal and bus on my unstable and I see that versions are the same in testing/unstable.
Wondering if I should reboot or suspend this evening
If you have some details on how this happened (dpkg.log,dpkg -l,..), you should open a bug report because it is very critical. Your report would be rerouted to the offending package if its not one of these 2.
Final news: I took about 2 hours to rebuild the ReiserFS trees of all the three partitions. The "/root" one ("/dev/hdb1") is totally gone (now I can mount it, but it contains the "lost+found" directory only, goodbye to all the rest), the "/usr" one ("/dev/hdb3") seems to be ok, even if reiserfsck had to correct a huge number of inodes and pointers (this means that many files could be corrupted), the "/var" one ("/dev/hdb4") is quite incomplete, and the "/var/log" directory is obviously gone... DAMN!!!
Now I have to begin a new installation... I think this is the good time to give Ubuntu 5.10 a try.
In the mean time, I really want to understand what's happened.
Debian users, please pay paranoid attention to dbus, expecially to its init script, and if you find something suspiscious about the unrecognized "--drop-privileges" parameter, break the time and try to solve it. Three partitions heavily damaged during boot process is a very bad thing.