[Solved] Pacman erratic behavior

jpsimard · 2013-07-03 00:46:46

At first, I thought I just suffered one of those odd events, but I just had basically the same problem again.

Short story, I do a full update on my laptop, and it appears as the update goes fine with no error messages. But, the problem shows when I try to shut down the system. The X server gets killed, the system drops to the command prompt, and hangs there, never shutting down and remains unresponsive. I have to force it to shut down. Rebooting works, but X refuses to start.

The first time, I retorgraded the kernel package, thinking that was it as it was just upgraded, and it did get the system to boot normally. Trying a new system update, pacman declared that both the kernel package and the wine package were corrupted. The things is both those packaged were installed in the update that screwed things up. How pacman let pass two corrupted packages is beyond me. Last time, retrograding the xf86-video-intel package got the system running again, but no package were declared as corrupted in the following attempt to update.

Looking at the pacman log, nothing sticks out except the resulting errors from the bad updates. To my relief, my desktop has had no such "event", just my laptop.

So I have no idea why this is happening and would like some opinions as to where to look. The only thing I could think of is that my hard drive is failing, but I have no real reason to think that other than this weird behavior. I don't want to file a bug against pacman if it's my hardware, but yet the laptop runs perfectly in it's daily use. Of course, I'm not ruling out that I might have screwed my laptop really bad, but it has been running in its current configuration for months without a hitch (after removing win7). The laptop is now fully updated and runs as if nothing happened.

So I'm not even close to being experienced enough to explain this crap away and fix it. Any opinions?

Last edited by jpsimard (2013-07-03 17:35:03)

Trilby · 2013-07-03 00:55:15

I'm confused. Is the only problem/symptom the failure to poweroff at the end of a shutdown? And has that symptom only been displayed a couple of times?

This is not to imply that it wouldn't still be a vaild problem, but you're opening up all sorts of suspicions about pacman, kernel packages, video drivers, etc, with no apparent data to justify any of those suspicions.

I suspect there is such data, but it's not coming across in your description. Remember, we only know what you elaborate for us. So can you clarify what are the clear and objective observations of the problem versus what parts are the suspicions/inferences that you have been lead to.

(edit: I'm not sure which is worse, my spelling or my typing. Between the two I must look completely iliterate)

Last edited by Trilby (2013-07-03 01:09:46)

karol · 2013-07-03 01:02:37

I'm really confused wrt what happened.
Please post the exact commands you used and exact output.

jpsimard wrote:

Trying a new system update, pacman declared that both the kernel package and the wine package were corrupted. The things is both those packaged were installed in the update that screwed things up. How pacman let pass two corrupted packages is beyond me.

Pacman checks if the packages it has just downloaded are OK and finds out that linux and wine are not OK.
If you installed wine during the new update, it should be a newer version than during the update that screwed things up i.e. a different package. Pacman didn't " let pass two corrupted packages" - it informed you they were corrupted, didn't it?

jpsimard · 2013-07-03 12:19:39

Yeah, sorry about not being very precise. It is kind of the true problem as there are so very few error messages, and nothing in the logs. I'll try to order things and be more precise.

1) A few days ago, I did an update with "pacman -Syu" which seemed to be entirely succesful. Since a kernel package was involved, I immidiatly rebooted.

2) The computer did not shut down properly. I had to force the shutdown manually by pressing the power button. Upon reboot, it did restart, but without X, which wouldn't run. Having just updated the kernel, I thought that perhaps retrograding it would get things back to normal. For some reason, it did and a reboot later, X started up.

3) Upon reboot, I looked at pacman.log to see if I missed something. The log did not contain any message of concern. So I tried a second update which then told me that the kernel package and wine, were corrupted. In other words, the packages that were just installed a minute ago were now corrupted and I learned of their state after pacman installed them just minutes before. I deleted the packages from the cache, and did "pacman -Syu" again, which seemed successful as before. I rebooted, and the laptop was working fine. I looked in dmesg and pacman.log, to find nothing of interest, so I chalked up the whole event as some random crap.

4) Skip a few days, and I did a "pacman -Syu", which again seemed to have completed succesfully. A reboot however showed a similar problem where the laptop didn't want to shutdown, leading to a forced shutdown on my part. Upon the next booting, X didn't start and was left with nothing but a login prompt as before.

5) Since I was left without X, and having just update xf86-video-intel, I took a chance and retrograded that package. It worked, and the computer rebooted to a normal state with X running, not that I know why it worked.

6) As before, I tried a new update, reinstalled xf86-video-intel and all was fine. I did forget to mention before that pacman complained that db.lck was present in /var/lib/pacman/ as if the update process was interrupted. I am perfectly sure that the update completed and that I didn't interrupt it. In any case, I had to "rm" db.lck.

Which in all, leaves just as clueless as you can get. No errors in the logs that I could find, and the faulty updates had no error message themselves. I'm left with corrupted packages and the presence of the db.lck in /var/lib/pacman as my only trace of any having anything gone wrong.

The only time I have had something this bizarre, I had bad ram, but then the computer was bizarre all day long. Right now, I'm thinking more around the hard drive. A bad drive would explain why packages were fine to install once, but were in a corrupted state later. It would (at least I think it would) explain why db.lck was present after pacman finished the update, as the drive might not have responded correctly to the deletion of db.lck. So I'll look for tools to test the drive, just to make sure. Maybe the drive is fine, but my ext4 partitions could be funny...

Mind you, I don't really expect a direct solution here, as I have nothing to go on as you have noticed from my first post. What I want is perhaps suggestions as to what I should look at, maybe something you think I missed. I want to find the solution myself, but I have so few ideas of where to look next past dmesg or pacman.log.

Last edited by jpsimard (2013-07-03 12:45:54)

hokasch · 2013-07-03 13:02:00

Hmm, obscure. Have you checked Xorg logs when it failed? Are you running low on disk space?

jpsimard · 2013-07-03 17:30:11

I just updated the laptop, and all went fine. I thought that it would be a good test to see if things were really screwed, since it would be 3 out of 3 updates that failed to execute correctly. But today, all is fine...

@Hokasch: No I didn't check xorg logs, so obsessed I was that the update broke things, I was just thinking about what to downgrade. I'm admittingly not alway to smartest tool in the shed. As for disk space, my smallest partition has 10GB free space, so nothing there to worry about.

Since I could update today without issue, I guess I'll just have to hope it won't happen again and leave well enough alone. Further, S.M.A.R.T. information for the disk, along with tests have delared the HDD as healthy, showing my theory as incorrect. At least I know my drive is good.

Thanks for your time guys. I'll just marked this as solved even if I still don't have any idea what happened.

Arch Linux

#1 2013-07-03 00:46:46

[Solved] Pacman erratic behavior

#2 2013-07-03 00:55:15

Re: [Solved] Pacman erratic behavior

#3 2013-07-03 01:02:37

Re: [Solved] Pacman erratic behavior

#4 2013-07-03 12:19:39

Re: [Solved] Pacman erratic behavior

#5 2013-07-03 13:02:00

Re: [Solved] Pacman erratic behavior

#6 2013-07-03 17:30:11

Re: [Solved] Pacman erratic behavior

Board footer