Bug 140124

Summary:	rcnetwork ignores eth0 during the first run
Product:	[openSUSE] SUSE Linux 10.1	Reporter:	Martin Vidner <mvidner>
Component:	Network	Assignee:	Christian Zoz <zoz>
Status:	VERIFIED FIXED	QA Contact:	E-mail List <qa-bugs>
Severity:	Major
Priority:	P5 - None	CC:	fmfischer, juergen.mell, pmladek, suse-beta
Version:	Beta 6
Target Milestone:	---
Hardware:	x86-64
OS:	Other
Whiteboard:
Found By:	Other	Services Priority:
Business Priority:		Blocker:	---
Marketing QA Status:	---	IT Deployment:	---
Attachments:	ifcfg-eth-id-00:50:fc:f7:de:d8 netdebug4 netdebug5

Description Martin Vidner 2005-12-19 12:40:16 UTC

When booting a system, rcnetwork ignores eth0. When run the second time, it is found and brought up. Combined with bug 135589, logging in is not possible, runlevel 1 must be used.
Logs follow.

Comment 1 Martin Vidner 2005-12-19 12:48:19 UTC

Created attachment 61354 [details]
ifcfg-eth-id-00:50:fc:f7:de:d8

Comment 2 Martin Vidner 2005-12-19 12:51:13 UTC

Created attachment 61355 [details]
netdebug4

I booted to runlevel 1.
This is the first run of "rcnetwork start -o debug >netdebug4 2>&1"
Note that eth0 exists in /sys/class/net but not in PHYSICAL_INTERFACES. /var/run/sysconfig/new-stamp-3 contains "renamed".

Comment 3 Martin Vidner 2005-12-19 12:54:07 UTC

Created attachment 61357 [details]
netdebug5

This is the second run: "rcnetwork start -o debug >netdebug5 2>&1"
/var/run/sysconfig/new-stamp-2 contains "unknown".

Comment 4 Martin Vidner 2006-02-08 09:16:45 UTC

Petr, I suppose this has started to work in the meanwile, right?

Comment 5 Christian Zoz 2006-02-13 09:34:29 UTC

*** Bug 140700 has been marked as a duplicate of this bug. ***

Comment 6 Christian Zoz 2006-02-13 09:35:15 UTC

*** Bug 146702 has been marked as a duplicate of this bug. ***

Comment 7 Christian Zoz 2006-02-13 09:35:58 UTC

*** Bug 148285 has been marked as a duplicate of this bug. ***

Comment 8 Christian Zoz 2006-02-13 09:38:53 UTC

There are still some problems at boottime. I have fixes for most issues, but i still need to test them. Stay tuned for next beta.

Comment 9 Christian Zoz 2006-02-13 09:40:56 UTC

*** Bug 119855 has been marked as a duplicate of this bug. ***

Comment 10 Christian Zoz 2006-02-13 15:35:30 UTC

Patches are in svn now and will be in beta4. Please test this, since all this is very prone to race conditions. Thus i cannot catch eb´very case myself.

It may already be tested internal with package from mbuild/oran-zoz-12.

Comment 11 Christian Zoz 2006-02-14 08:59:08 UTC

Still not fixed completely, sigh.

Comment 12 Christian Zoz 2006-02-14 19:02:40 UTC

There were still some wrong returnvalues which made rcnetwork think that setup of interface did succeed desipte it had still to wait.

It had worked in some cases even with wrong returnvalues, but additionally there was no /dev/shm/sysconfig early enough. So rename_netiface could not write renaming state and rcnetwork had to guess it. This is also fixed now. See bug 150828.

Fixes will be into beta4.

Comment 13 Christian Zoz 2006-02-14 19:39:08 UTC

Submitted fixed package.

You can also use packages from /mounts/mbuild/fix-zoz-2

Comment 14 Frank-Michael Fischer 2006-03-04 08:35:33 UTC

Now with Beta 6 situation got worse:

"rcnetwork status" says: running
but network is not there.

After "rcnetwork restart" we get: running
but network is still not there and the DHCP client (unlike in earlier betas) does not get an ip address.

My workaround: Set a static address and do a "rcnetwork restart".

Comment 15 Christian Zoz 2006-03-30 14:06:13 UTC

Is that still valid?

If yes, please set /etc/sysconfig/network/config:DEBUG=yes and then send me the complete output from network boot script (from /var/log/boot.msg).

Comment 16 Christian Zoz 2006-04-12 12:10:18 UTC

No reaction for two weeks.
Further no other reoprts regarding this and i did not have this problem on my machines.

Back to 'fixed'.