final hoary testing report

[James Troup]

Hi,

This is a roundup of the last couple of days of testing I did - sorry
for the delay in posting - I got distracted by fighting firmware and
various other things on Thu/Fri.  I'll try and turn all the
outstanding issues from this and the other reports into bugs over the
next couple of days.

-- 
James

================================================================================

Install (i386) test on HP Proliant DL380 G3
-------------------------------------------

 + All good.

Install (amd64) test on HP Proliant DL380 G4
--------------------------------------------

 + All good.

Live (i386) test on IBM eSeries 325
-----------------------------------

 - failed to boot unless 'hpet=disable' passed, hung after "Using hpet
   for high-res timesource"

 + Otherwise, all good

Netboot Install (i386) test on HP Proliant DL140
------------------------------------------------

 - Our netboot docs aren't great.  There's on page on the wiki
   repeated 3 times which (no disrepect to the author) is an
   incomprehensible mess if all you want to do is simply PXE-boot a
   server.  The Debian docs are better but buggy too.

 - Why do we duplicate netboot.tar.gz and it's contents in
   installer-i386/current/images/netboot/?  It's very confusing [and
   we aren't consistent about it across architectures]
 
 - Prompts me for a mirror to choose?  In both first and second stage too...

 - Warns me about ejecting CD-ROM etc.  Even qualified as it is, it
   seems pointless.  It should be able to know I'm not using a CD-ROM.

 - 'server' didn't tie over to second stage?  Still tried to install
   'mdetect' etc.

 + Otherwise, all good
 
Upgrade (amd64) test
--------------------

 - lib32gcc file overwrite errror

| Unpacking lib32gcc1 (from .../lib32gcc1_4.0-0pre6ubuntu3_amd64.deb) ...
| dpkg: error processing /var/cache/apt/archives/lib32gcc1_4.0-0pre6ubuntu3_amd64.deb (--unpack):
|  trying to overwrite `/lib32/libgcc_s.so.1', which is also in package ia32-libs-openoffice.org
| dpkg-deb: subprocess paste killed by signal (Broken pipe)

 . Expected bogus conf file questions

 + Otherwise, all good

CD Install (amd64) test on IBM eSeries 325
------------------------------------------

 + All good

Netboot Install (amd64) test on IBM eSeries 325
------------------------------------------------

 . Same problems as i386 netboot

 + Otherwise, all good

Upgrade (i386) test
-------------------

| Unpacking replacement perl-base ...
| dpkg: error processing /var/cache/apt/archives/perl-base_5.8.4-6_i386.deb (--unpack):
|  trying to overwrite `/usr/lib/perl/5.8', which is also in package liblockfile-simple-perl
| Errors were encountered while processing:
|  /var/cache/apt/archives/perl-base_5.8.4-6_i386.deb
| E: Sub-process /usr/bin/dpkg returned an error code (1)

 . Expected bogus conf file questions

 + Otherwise, all good

Netboot (powerpc) test on Apple Xserve G5
-----------------------------------------

 - no netboot dir on archive.u.c still

 - didn't matter much anyway - OF is broken on these stupid Xserves
   and can't get yaboot.conf, see

   http://lists.penguinppc.org/yaboot-users/2005/yaboot-users-200501/msg00013.html

Install (i386) test on IBM xSeries 345
--------------------------------------

 - LONG blank screen pause after creating FS on .5TB drive 

 + All good.

(Revisited) Install (powerpc) test on Apple Xserve G5
-----------------------------------------------------

 - At the 'Apt configuration' point near end of stage 1; apt-cdrom
   type output spams the screen. [Colin fixed this on Friday, I think]

ia64 upgrade test
-----------------

 o We should note in upgrade-from-debian notes for ia64 that Debian
   uses /efi/debian and Ubuntu uses /efi/ubuntu; since the dirname is
   hardcoded into the boot option, people probably won't realise
   they're using the Debian kernel still.

 o VFAT appears to be SNAFU in our 2.6.10?  Simply mv-ing a directory
   off of a VFAT partition seems to corrupt it.

 o elilo silently hangs if dosfsck finds problems with the boot
   partition.  score.

 o Otherwise ok

Netboot Install (ia64) test on HP Integrity RX1600
--------------------------------------------------

 = Had LOTS of problems with this; it turns out the firmware on
   this box (1.10) has an incredibly evil bug where it's tftp client
   only works once and will fail in non-obvious ways on each
   subsequent attempt until you cold reset.  'Upgrade your firmware if
   you're trying to netboot' should be in big flashing writing
   somewhere in our ia64 install docs.  

   [Many thanks to Dann Frazier for helping me to diagnose this.]

| The proxy information should be given in the standard form of 
| "http://[[user][:pass]@]host[:port]/"
| HTTP proxy information (blank for none):
| Prompt: '?' for help> 
| Checking the Ubuntu archive mirror  
| Choose a mirror of the Ubuntu archive
| 
| !! ERROR: Bad archive mirror
| 
| The specified Ubuntu archive mirror is either not available, or does not have a
| valid Release file on it. Please try a different mirror.
| [Press enter to continue]

  a) Why ask me for a proxy?, b) what mirror? ... 

 - can't get any further than this...

================================================================================