Bug 447971

This is not a blocker according to the definition.

Please attach the YaST log files as explained at http://en.opensuse.org/Bugs/YaST as well as the output of hwinfo.

Created attachment 254567 [details]
archive created from /var/log/YaST2/y2log* as currently found on my system

This archive contains the following files as currently seen (after failed upgrade then downgrade) in /var/log/YaST2/
y2log
y2log-1.gz
y2log-2.gz
y2log-3.gz
y2log-4
y2log.SuSEconfig
y2logMount
y2logRPM
y2log_bootloader
y2logmkinitrd
I don't know which ones (if any) may be relevant to the present bug.

The hardware detected by YaST2 as run under 11.0 (the working system after downgrade, not the failed system installed by upgrade, but the hardware is of course the same -- only /dev/sda instead of /dev/hda, /dev/sr0 instead of /dev/hdc, etc.) is found as attachment 254549 [details] with the other bug I filed about this failed upgrade (bug 447970).

About making this a blocker: IMHO a system upgrade which leaves me with no Internet access is totally unusable and must be solved before final release. If you don't agree, well, I suppose the last word rests with you; but I'm not going to install 11.1 final if it ships with this bug.

P.S. The attachment and comment above provide the required information, I think. Please set NEEDINFO again if there is more I can tell you.

Regarding blocker: Please read the definitions at http://en.opensuse.org/Bugs/Definitions.  There are other ways to mark a bug as ship stopper.

(In reply to comment #4 from Andreas Jaeger)
> Regarding blocker: Please read the definitions at
> http://en.opensuse.org/Bugs/Definitions.  There are other ways to mark a bug as
> ship stopper.
> 

I just read it:

    *  Prevents developers or testers from performing their jobs. Impacts the development process.
    * (Documentation) Key documentation is missing for critical testing and review. 

That bug sure "prevents" me (as an occasional voluntary tester) from "performing my job" (i.e., from live-testing that version of the openSUSE distribution). If I can't get an Internet connection, I cannot even report a bug, not to mention all the other things I cannot do which force me to downgrade. (Bug 447970 is pretty serious too but at least I found a workaround, albeit kludgy.)

I finally found the SHIP_STOPPER flag, but I don't know whom to request it from. Feel free to fill in the flag's requestee.

Michal, please check the logs. I need to know if this is a generic problem or a config problem / hardware problem.

IMHO something wrong with hwinfo (maybe driver missing?)

2008-11-22 18:05:35 <1> linux(10402) [YCP] clients/hwinfo.ycp:174 Probing dsl (DSL)...
2008-11-22 18:05:35 <3> linux(10402) [Interpreter] clients/hwinfo.ycp:184 SCR::Dir() failed


Ludwig, any idea?

Do you have 11.0 and 11.1 installed in parallel? If so could you run pppoe-discovery in both installations? It should find your modem. If it only doesn't work on 11.1 it's probably a network driver problem.

pppoe-discovery and a manually started pppd work fine for me under 11.1Beta5, but when I try to "ifup dsl0" pppd gets started and exits after 30 seconds, leaving "No response to PAP authenticate-requests" in the syslog.

Still investigating...

(In reply to comment #8 from Ludwig Nussel)
> Do you have 11.0 and 11.1 installed in parallel? [...]

No I don't. I tried to upgrade from 11.0 to 11.1, found bug 447970 (and a kludgy workaround for it) then this bug (and no workaround) so I had to downgrade. At no time did I have more than one Linux system installed on my hard disk.

FYI, I'm pasting below the output from ifconfig under 11.0. The ethernet card uses static addressing, the DSL interface dynamic addressing.

dsl0      Link encap:Point-to-Point Protocol  
          inet addr:81.245.59.121  P-t-P:81.245.59.1  Mask:255.255.255.255
          UP POINTOPOINT RUNNING NOARP MULTICAST  MTU:1492  Metric:1
          RX packets:93316 errors:0 dropped:0 overruns:0 frame:0
          TX packets:98257 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:3 
          RX bytes:56616462 (53.9 Mb)  TX bytes:19440865 (18.5 Mb)

eth0      Link encap:Ethernet  HWaddr 00:14:2A:62:74:64  
          inet addr:10.0.1.0  Bcast:10.0.1.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:396885 errors:0 dropped:0 overruns:0 frame:0
          TX packets:298601 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:349532035 (333.3 Mb)  TX bytes:49976415 (47.6 Mb)
          Interrupt:19 Base address:0xe400 

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:69493 errors:0 dropped:0 overruns:0 frame:0
          TX packets:69493 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:3532942 (3.3 Mb)  TX bytes:3532942 (3.3 Mb)

...and here is the output of pppoe-discovery, but only on 11.0 since I don't have 11.1 installed.

Access-Concentrator: 90084090400334-BAS-PAILLE2
--------------------------------------------------
AC-Ethernet-Address: 00:02:3b:02:9c:97


Oh, one more thing: You may want to check the kludgy bootup workaround in bug 447970. IIUC, it means that in order to run the 11.1 system I had to run init with the /etc/inittab on the HD from the kernel on the DVD. Don't know if it makes a difference.

Well, next to impossible to reproduce then. We couldn't reproduce the problem Reinhard saw either. PPPoE works in principle so there is no generic problem with that kind of connection.

Do you still have /var/log/messages with the entries when you tried to dial in on 11.1 (if so please attach)? Also the /etc/sysconfig/ifcfg-* files would be interesting.

If you want to verify that your ADSL setup works on 11.1 you could give the the live-cd a try.

ok, doesn't sound like a ship stopper

Created attachment 254812 [details]
extract from /var/log/messages

(In reply to comment #11 from Ludwig Nussel)
[...]
> Do you still have /var/log/messages with the entries when you tried to dial in
> on 11.1 (if so please attach)? Also the /etc/sysconfig/ifcfg-* files would be
> interesting.

I found /var/log/messages and extracted the part concerning my attempts at running the 11.1-Beta5 system. I added two comments starting with a line of dashes (to distinguish them from the "date" starting all "standard" lines) for the upgrade and downgrade. The first 11.0 session after the downgrade is also included.

There is no /etc/sysconfig/ifcfg-* but I'll attach /etc/sysconfig/network/ifcfg-* as a further attachment.

> 
> If you want to verify that your ADSL setup works on 11.1 you could give the the
> live-cd a try.
> 

OK, I haven't done it yet but I'll do it sometime soon; probably tomorrow (Tuesday) or maybe the day after. (I'm on European time.)

Created attachment 254813 [details]
/etc/sysconfig/network/ifcfg-dsl0

Created attachment 254815 [details]
/etc/sysconfig/network/ifcfg-eth0

I think the essential part is this:
--- snip ---
pppd[12131]: Couldn't open the /dev/ppp device: No such device or address
pppd[12131]: Linux kernel does not support PPPoE -- are you running 2.4.x?
pppd[12131]: Exit.
--- snap ---

For whatever reason, the ppp kernel modules don't seem to be loaded on your system, or at least the corresponding device node doesn't get created.

That because the system was booted with a different kernel (the one from the DVD) than the one that was installed.

Just an unrelated side note on your ifcfg-eth0 file:

--- snip ---
IPADDR='10.0.1.0'
NETMASK='255.255.255.0'
NETWORK='10.0.1.0'
--- snap ---

You are using the network address as your machine's IP address.
This is not a good idea.

Neither the network address (x.y.z.0 in a class C network) nor the broadcast address (x.y.z.255 in a class C network) should be used as the IP address for a machine.

(In reply to comment #18 from Reinhard Max)
> Just an unrelated side note on your ifcfg-eth0 file:
> 
> --- snip ---
> IPADDR='10.0.1.0'
> NETMASK='255.255.255.0'
> NETWORK='10.0.1.0'
> --- snap ---
> 
> You are using the network address as your machine's IP address.
> This is not a good idea.
> 
> Neither the network address (x.y.z.0 in a class C network) nor the broadcast
> address (x.y.z.255 in a class C network) should be used as the IP address for a
> machine.
> 

All I can say is it works, so, as it's out of my line of competence, "since it ain't broke, I don't fix it". That network is between my computer and the DSL device, which has no address switches on it.

(In reply to comment #19 from Tony Mechelynck)
> All I can say is it works, so, as it's out of my line of competence, "since it
> ain't broke, I don't fix it". That network is between my computer and the DSL
> device, which has no address switches on it.

Ah - in that case it indeed doesn't hurt, as an interface that is only being used for PPPoE will never see actual IP traffic, and therefore doesn't even need to have an IP address at all. But on a normal LAN you most probably would have run into problems with an IP address that is identical to the network address, so I thought it would be good to warn you about it.

(In reply to comment #20 from Reinhard Max)
> (In reply to comment #19 from Tony Mechelynck)
> > All I can say is it works, so, as it's out of my line of competence, "since it
> > ain't broke, I don't fix it". That network is between my computer and the DSL
> > device, which has no address switches on it.
> 
> Ah - in that case it indeed doesn't hurt, as an interface that is only being
> used for PPPoE will never see actual IP traffic, and therefore doesn't even
> need to have an IP address at all. But on a normal LAN you most probably would
> have run into problems with an IP address that is identical to the network
> address, so I thought it would be good to warn you about it.
> 

Well, thanks for the warning. If ever I use Ethernet for an actual LAN (but I don't think I will, at least not anytime soon), I'll remember this -- I hope. :-)