Bug 289553

Created attachment 149634 [details]
bug289553.tar.bz2

Missing a bug description, telling me what really happened.

What have you done? 

Installed patch/packages/?? XYZ?

yast installs -64bit.ppc.rpm packages on systems that had no glibc-64bit.ppc.rpm installed during fresh install. This must not happen. Trouble with the patch meta files.

Please provide a _detailed_ description how to reproduce this!

It seems you have used YaST.
Did you run an update using YaST Online Update? (Y/n)

Did you install an package using YaST package management (sw_single) (y/N)?

If you installed patches, which patches was selected?


I need these answer for deciding if it is a patch or a update stack issue.

Created attachment 149852 [details]
bug289553-kiwi.tar.bz2

I just click next next next next during a fresh install in second stage.

new logfiles from kiwi.suse.de after a fresh install.

The is no patch for glibc nor glibc-64bit in the repository.

Need to find out which packages/patch requires glibc-64bit, to do this I need a full logging enable. Is it possible to install the machine again and NOT updating during installation or enable full logging before online update starts (people from YaST team might know how/if this is possible).

How many foo-64bit pacakges are available for 10.2?
Each one gets installed unconditionally if foo is installed and thus gets selected for update.
Each foo-64bit.rpm requires glibc-64bit. But thats not the point.

If foo.rpm is installed and and yast decides that there is an update for foo available, yast must not select foo-64bit.rpm unless foo-64bit.rpm is already installed.

I dont think its a yast bug.

Olaf, please, provide a log with ZYPP_FULLLOG=1 

Created attachment 152404 [details]
y2logs.tar.bz2

Ouch !
It was an explicit request by the PPC team to switch to -ppc64 package where applicable in the SLE10GA -> SP1 update.

Now you're asking not to switch ??

Please define _clear_ conditions when a switch should happen and when not.

this is not about foo.ppc64.rpm, but about foo-64bit.rpm.
There are no .ppc64.rpm packages on 10.2, beside mysql and stuff.

so what exactly triggers the installation of bind-64bit.rpm?
Note that this does not happen every time. 

Ok, then I misunderstood the thread.

Can you please describe in words, what is currently happening and what the expected behaviour would be ?

see comment #3 and the question in comment #12

I wouldn't call this answer as "describe in words, what is currently happening and what the expected behaviour would be ?", so I'm still guessing what happens and what might be expected :-(

bind.ppc.rpm is installed and upgraded to bind-64bit.ppc.rpm during online-update in 10.2

If this must not happen and the -64bit package was not available during initial install, is also must not be available in the update repository

-> Maintenance

If Olaf can't describe the problem, I think INVALID is fine

what exactly is hard to understand in comment #3 and the question in comment #12?

a foo-64bit.ppc.rpm is very differend from a foo.ppc64.rpm.
The latter can not be installed on a 32bit system.
In fact, no *.ppc64.rpm can be installed on a 32bit system.
And because a *.ppcc64.rpm cant be installed on a 32bit system, yast should not pull in *-64bit.ppc.rpm packages.

So once again:

What exactly triggers the installation of *-64bit.ppc.rpms during an online update?
I think its safe to ignore all request for logs if noone is able to parse them...

Stefan can not parse yast logs either.

Its getting boring. Stefen clearly cant do anything with libzypp internals.

All the damn info is in comment #3, comment #12 and comment #18
If the logs with ZYPP_FULLLOG=1 are incomplete, what else can be done to increase verbosity?

Olaf! Stop doing bugzilla ping pongs!

I think Stefan can very well speak for himself.

Its clearly not ok to tolerate your childish games.

I will check with 10.3 GA if the bug is still there.

mass reopening all 10.2 LATER+REMIND bugs.

close all 10.2 LATER/REMIND bugs as WONTFIX. Reopen yourself if you still plan to work on it.