[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: What happened to debian - does "stable" keep having any meaning?



Hello,

Jochen Schulz wrote:
> Geronimo:
> > Jochen Schulz wrote:
> >> Geronimo:
> >>> the last update of debian broke my system completely!
> >> 
> >> I am very sorry for your wasted time and loss of data. I see why you
> >> need to let off steam.
> > 
> > Thank you very much! - usually I'm not that coarse.
> 
> That's ok. I am glad my mail didn't anger you even more. :)

No person caused any anger to me.

I got angry and frustrated by the fact, that the same issue, I reported before 
squeeze becoming stable now came back to break my stable system.

I confess, that I might have glorified debian comunity too much - and may be, 
my anger arised about the difference between my thinking and reality ...

> - You have five disks (SSD or hard disks, shouldn't matter):
> 
>   / on sda1, ext4
>   /boot on sda2, ext2
>   swap on sdc1
>   /usr on sdc2, ext4 (btw, it's "UNIX system ressources", not "user")
>   another swap on sdd2
>   /var on sdd2, ext4
> 
>   sdb and sde appear to be unused with respect to the squeeze system
>   (You may use them with squeeze, but they don't hold any
>   system-relevant data.)

Not quite.

I have four "system disks" and a hotswap backplane with four "user disks", 
which are powered on demand, so system may have four to eight disks.
The system disks are attached to the main board and are permanently powered.
The backplane-disks are disks exchanged with other machines and they don't 
contain system relevant data.
Each disk of a pool of exchangeable disks is labeled to the same name, so I 
can mount different disks from the same pool without changing fstab.

sdb is a System disk with data, sde is a user disk with data.

What happened:

After the system was broken, I was stuck to the cd/dvd I had, which is the 
debian installer disk 6.0 stable and an ubuntu 10.10 live. The debian 
installer disk is my first choice for rescue operations, but I did not succeed 
to bring the system into play.
So I startet the ubuntu live, to be able to google for recovery tips.
I accidently found a script that tells me what grub sees from the disks.
Sadly I placed that in virtual storage so I don't have the results and I don't 
remember the name of the script.

... anyway:
The script told me, that every disk contained a grub installation in mbr, so I 
thought, might be a problem for grub and searched for a way to remove grub 
from mbr.
The only way, I found, was a dd sequence that wipes out the first block.

What I did not know before: that dd sequence removes all partition data.
I executed that dd sequence against 7 of my disks and after reboot I was so 
*frightened* about the damage I did.

So I searched for a way to recover partition informations. That leaded me to 
testdisk, which worked fine for the SSDs. Therefore 3 of my system disks are ok 
again. The rest of the disks is damaged.

I run testdisk on the VelociRaptor too, and let it write the found data, but 
the result was wrong.
So my current state is: one disk with wrong partition table and three disks 
with wiped out partition table.
All four disks contain valid data with usage of 50 to 90 percent with sizes 
from 150GB to 2TB.
 
> - Some of these disks are attached to a secondary SATA controller
>   (RocketRAID 230x).
> 
> - Other disks are attached to the mainboard's (GA880GM-UD2H) controller.
> 
> - Your setup worked fine even after you upgraded from lenny to squeeze.

Yep.

> - You recently upgraded to the next point release and in the process
>   were asked to reboot. After that, your system didn't boot. Btw, I am
>   curious what exactly triggered the reboot warning. I cannot remember
>   having seen that.

I'm using KDE as desktop and the message arised from KDE. AFAIK it was the 
same, when I used gnome desktop.

> - Which disks are connected to which controller?

As told above, the "system disks" are attached to the main board. The system 
disks are the three SSDs and a WD VelociRaptor (sdb) 

> - How did you upgrade to 6.0.1?

I have a bash alias for this task:
alias dist-upgrade='apt-get update && apt-get dist-upgrade && apt-get clean'

That get's executed (nearly) every day before I start work.
excerpt from dpkg.log is attached.

> - How often did you reboot the system after the upgrade from lenny to
>   squeeze? I am not interested in the exact number, I just want to
>   ensure you did it at all. :)

May be 400 times? - Not sure.
 
> - How did you configure the secondary controller? AHCI?

You mean the secondary controller from mainboard? - Yes, AHCI.

> - Have you tried shuffling your disks around? Does the system respond
>   differently depending on the controller for / or /boot?

The system shuffles the disks around when I switch a disk from backplane on/off 
or enter a USB-stick, so yes - I'm very used to reorder disks at BIOS.
When the wrong disk is first, system start is not possible.

> > So, may be you can guide me to a better use case, when the system is
> > broken and the machine does not work any more.
> 
> Well, I am not too proficient in these things either. I throw my
> problems at Google like everyone else. :) But installing a new system on
> top of an old one wasn't a good idea when I still used Windows 95 (good
> riddance) and it still isn't a good idea today. /usr may contain
> binaries which the system doesn't know about and thus don't receive
> security updates. And /var contains system-specific data like dpkg's
> database that you don't want to re-use on a new installation.

So how can I recover a broken system?

/usr and /var both have a "local" directory where I put several directories, 
which then get mounted to the right location via bind option. So I don't want 
these drives get formatted.
 
> But anyway: if you have a reliable way of crashing the installer, use
> reportbug (pseudo-package: debian-installer) to make the d-i team aware
> of the problem. 

Thanks for that hint.
I'll try to remember, when I have my next sparetime.

> Running through the gun club yelling fire will get you attention, but
> not the assistance you wish.

You're totally right. 
I asked for assistance about 6 weeks ago - so this time the start of this 
thread was no request for help. That changed during thread lifetime.

> A tirade without any useful information on the actual problem does
> offend (especially around midnight).

Don't know if you can imagine the amount of my disapointment and violation 
with the last adventure. 
The same happening did not rise any anger, when squeeze was testing. Then I 
had to doom myself for using testing.
I was so sure, that this will never happen with debian stable ...

So sorry for all my f*#ing words.


kind regards

Gero

Attachment: dpkg.log.tar.bz2
Description: application/bzip-compressed-tar


Reply to: