Returning from hibernate doesn't start the INTERFACES (network) on Feisty.

Hello Peter,

Can you attach a diff showing how the file was before, and what the solution is? This makes it easier to see exactly what needs uploading to fix the issue.

Here is the patch with diff -Naur

Two quick comments: 1) the attached diff doesn't have the necessary ` , so INTERFACES won't get set properly, but more importantly,

2) INTERFACES is set in exactly this way in /etc/acpi/suspend.d/55-down-interfaces.sh

All scripts in suspend.d should be being called on the way down; if they are not, then that is the source of the bug: As part of going down, sufficient state is saved to restore the system properly on the way back up (at least that's the theory). In general, the resume scripts should not be setting variables, they should be using those set during suspend.

Peter, if you can, please track down why INTERFACES isn't being set on the way down. Thanks!

You are right and right. I forgot the ` when I cut-and-paste in joe. I should have used scp, as I was using a remote machine for the diff. The second point also is correct. I removed the setting in the resume.d for the INTERFACES and I asked the scripts to save the state of INTERFACES (simple placed an echo $INTERFACES >/tmp/ifdown respective ifup) and it looks like you said. Now, I don't know anymore what caused the problem with restarting the interfaces for resume? Complete puzzled. I will close the bug and if I get some more data, I will return with the results. I also did an upgrade to other modules, including the restricted-modules (I am using an intel ipw3945 network card that might have had some issues).

Anyway, thanks for the help...

Cheers,
Peter

Seems like it is working, in fact...maybe the bug is somewhere else...

On Monday 05 March 2007 23:04:23 Peter Whittaker wrote:
> Peter, if you can, please track down why INTERFACES isn't being set on
> the way down. Thanks!

I've been looking more into this issue and found out some things that you
might consider into the acpi-support package. What I found is that the dbus
is not reset after return from hibernate. This is quite an important thing
and can be done easily by just adding a script in resume.d
like: /etc/init.d/dbus restart. If you don't do this restart, for example, if
the user changes from using eth0 (wired) and returns into using eth1
(wireless), it might be that the NetworkManager will not be able to detect
the interfaces. Are you aware of this issue with the NetworkManager? I can
explain more the issue if you need...but what are your thoughts on this
topic?

Thanks,
Peter

--
                                  o
 Dr. Peter Antoniac (theSeinfeld) I~./
 http://antoniac.info/ Z\ T