Manual intervention required

[pepedog]

Details here, /bin, /sbin, and /usr/sbin go and become symlinks to /usr/bin
https://www.archlinux.org/
Basically, for me, I had to-

Code: Select all

pacman -R localepurge
pacman -Syu --ignore filesystem,bash
pacman -S bash
pacman -Su

If final step fails, are there any files in /usr/sbin ?
if not then delete folder
If there is the pacman -Qo /usr/sbin/thefile and get packagename. then pacman -R packagename

There is another unrelated problem, do this before reboot

Code: Select all

pacman -U http://us.mirror.archlinuxarm.org/armv7h/core/netctl-1.1-1-any.pkg.tar.xz

It's an "any" package
This should be fixed soon

[danno]

Thanks. I had been struggling with this since early morning.

Checking /usr/sbin and removing the package 'localepurge' solved my issue.

Also, the netctl package pacman downloaded earlier this morning was corrupt. After removing it from cache, pacman re-downloaded the file and all went well. I'm not sure what the issue was, but I'm guessing it's been fixed (?).

EDIT: Scratch that. After rebooting, I'm left with no network.

In dmesg, I have the following:

Code: Select all

systemd[1]: Cannot add dependency job for unit netctl-ifplugd@eth0.service, ignoring: Unit netctl-ifplugd@eth0.service failed to load: No such file or directory. See system logs and 'systemctl status netctl-ifplugd@eth0.service' for details.

Also,

Code: Select all

# systemctl status netctl-ifplugd@eth0.service
netctl-ifplugd@eth0.service
      Loaded: error (Reason: No such file or directory)
      Active: inactive (dead)

Dec 31 18:00:20 raspberrypi ifplugd[118]: client: Started network profile 'ethernet-dhcp'
Dec 31 18:00:20 raspberrypi ifplugd[118]: Program executed successfully.
Dec 31 18:00:22 raspberrypi dhcpd[240]: eth0: sending IPv6 Router Solicitation
Dec 31 18:00:26 raspberrypi dhcpd[240]: eth0: sending IPv6 Router Solicitation
Dec 31 18:00:22 raspberrypi dhcpd[240]: eth0: No IPv6 Routers available
Jun 04 08:43:55 raspberrypi dhcpd[240]: eth0: renewing lease of 192.168.1.21
<....snip>

EDIT #2: As best I could tell, the netctl package that got installed installed using 'pacman -Syu --ignore filesystem,bash' placed files in the wrong location. I removed it using pacman -R, but still needed to manually delete some files it placed in /. After that I downloaded the netctl package from the link above using a seperate computer, copied it to the pi, and installed. All is working now.

[sdjf]

Wow, I think I am waiting until all this gets worked out before updating again. But looks like it will not get fixed except with a whole lot of work. I wish they wouldn't keep breaking things, at least I can update a backup and when and if that gets functional, then change to using it.

I would think that the updates would have routines built in that automated these changes, are they supposed to put them in there so if I wait for another release, the system would fix itself properly?

[pepedog]

netctl has been fixed now. I have upgraded a raspberry trimslice, a trimslice, dockstar, and a zyxel now.
A new image is coming soon

[sdjf]

netctl has been fixed now.

Excellent news, thank you!

A new image is coming soon

I think I am passing on downloading that one, the last one took three downloads over four days to get an image that was not corrupt, and a couple attempts to get a place for unzipping it where it would fit.

Also am still testing a lot of GUI apps that are a lot of work to reinstall on a new card, so the card I have a new image on is to test procedures I am having trouble with on my main SD cards, like to see how Pacman really works when there are no unresolvable package conflicts when trying to upgrade, and seeing how to get my ethernet working on a vanilla system.

And it is a good backup card in case the other has serious problems.

[grrrue]

Hi, im having this issue on a rpi with a freshly installed image from the download section.

removing localpurge worked for me but on the second step, i get dozens of these:

Code: Select all

resolving dependencies...
warning: ignoring package filesystem-2013.05-2
warning: cannot resolve "filesystem>=2013.01", a dependency of "glibc"

Then im presented with a huge list of packages that cannot be updated due to dependencies of glibc. Regardless pacman continues to install a few packages like kernel images etc..
Trying to continue with

Code: Select all

pacman -S bash

leaves me with

Code: Select all

(1/1) upgrading bash                                                                                                                           
call to execv failed (No such file or directory)
error: command failed to execute correctly

Not sure where to dig, since i already tried lots of stuff and couldn't come up with any solution that doesn't bite me in the *ss in some other place.

Thanks

[pepedog]

Try
pacman -Syu --ignore filesystem,bash,glibc
pacman -S bash
pacman -Su

I had this on one today, can't remember the steps I took. When was the last time you updated? Or is this the Feb 13 image?

[grrrue]

I'll try that. Yes i was using the image from February. Shall i try the new one? I suppose there would be much less packages marked for upgrading, hence less trouble doing so?

[pepedog]

It's worth the experience upgrading, if you trash it then get the new image.
If you see that only unimportant wiles are in the 3 directories, copy to /usr/sbin and then delete them.
Don't reboot without completing, files in /sbin symlink first thing sought on boot

[grrrue]

I indeed tried the new image and also

Code: Select all

--ignore filesystem, bash, glibc

Everything went fine so far... Should i proceed with bash, glibc or filesystem?

[grrrue]

*doh* should get some sleep and read everything. Gonna try bash and then -Syu
Thank you very much

[pepedog]

Do bash, filesystem and glibc are done together.
Don't do Syu as there might be more packages, just Su to finish off

[gaggio]

Morning guys,
I think i screwed something up during this update: after the bash update I can't login anymore... Before rebooting I noticed that webmin couldn't execute any command on the system with an error message like "can't find /sbin/bash"

Is there any way to set it right, maybe putting the SD card into another machine and modifying the bash path?

Thanks for your help,
Federico

[pepedog]

Yes, something like move files from /bin /sbin /usr/sbin
Then delete those directories
Then make the 3 symlinks to point to /usr/bin

[gaggio]

Thanks pepedog,
Just did what you suggested, the device seems to boot now, with the LEDs on correctly. Still can't login via SSH (no response from host). I'll try to connect a keyboard and display to see what's going on.

Thank you

[gaggio]

Finally connected to a display, I can just see the following errors:

Code: Select all

sh: cannot set terminal process group (-1): Inappropriate ioctl for device
sh: no job control in this shell
sh-4.2#

Now the last line is indeed a command prompt, where I can actually input commands with root rights, even if I didn't log in!

Thanks again

[sdjf]

How about running route and ifconfig to see whether any networks are up?

[gaggio]

I finally managed to get my hands on the rasp:

The output of ifconfig is:

Code: Select all

Warning: cannot open /proc/net/dev (no such file or directory). Limited output.

While route -n gives me:

Code: Select all

/proc/net/route: no such file or directory
INET (IPv4) not configured in this system.

Thanks for helping me out

[sdjf]

How about running the following to make sure your symlinks were entered properly?

ls -l /bin /sbin /usr/sbin

Also try the following. Gosh, also make sure /proc shows before looking for contents of subdirectories.

ls -l /proc /proc/dev

and how about the following to see how you are logged in?

whoami

[bluknight]

I had to do the following to get back my internet:
1. Download and rewrite to SD the latest RP image of June 15
2. On boot, no internet available so I do as root:
>cp /etc/netctl/examples/ethernet-dhcp /etc/netctl/eth0
3. Edit /etc/netctl/eth0 and choose iVp6 autoconfiguration (don't know why but this works for me)
4. As root:
>netctl start eth0
(but still not connected so I do)
>dhcpcd eth0
(network now connected)
>netctl enable eth0
Now I can update with pacman -Syu

[pepedog]

You didn't need to do any of that, netctl list would have shown
ethernet-dhcp was there already enabled by netctl enable ethernet-dhcp
We could investigate here, is your serial number all zeros? do cat /proc/cpuinfo
Next we could see if taking netctl out of the picture helps
netctl disable ethernet-dhcp
netctl enable eth0 (you added that)
systemctl enable dhcpcd@eth0
sync, reboot, and see if it come up. If not try dhcpcd eth0 (which does not use that profile you copied)
We need to find out what's causing this, if it's a matter of timing, modules loading, etc

[bluknight]

netctl disable ethernet-dhcp
netctl enable eth0 (you added that)
systemctl enable dhcpcd@eth0
sync, reboot, and see if it come up. If not try dhcpcd eth0 (which does not use that profile you copied)
We need to find out what's causing this, if it's a matter of timing, modules loading, etc

Thanks the systemctl brings up my network each time I boot without having to do dhcpcd, So my profile was ethernet-dhcp and not eth0. My seriel number: 00000000212d2d4f
But I was sure the first time I install netctl start eth0 failed. Now I know.

[pepedog]

Some commands use interface, some profile
https://wiki.archlinux.org/index.php/Netctl