Bug#587887: linux-base: please suppress useless debconf messages on first install
Hi Ben,
for the record, this workaround works for me: build a custom udeb and put the
following into a file in /usr/lib/post-base-installer.d/$some_filename
# work around # 587887
grep -v cdrom /target/etc/fstab /target/etc/fstab.new
mv /target/etc/fstab /target/etc/fstab.bak
mv /target/etc/fstab.new /target/etc/fstab
# avoid warning that no boatloader is installed by installing one early
chroot /target apt-get -y install grub
mkdir -p /target/boot/grub
chroot /target update-grub
Preseeding that warning about no bootloader didnt work. I used:
linux-base linux-base/disk-id-manual-boot-loader error
On Freitag, 27. August 2010, Ben Hutchings wrote:
I think we've discussed this before.
Yes. In this report.
linux-base is supposed to be quiet
during a fresh system installation. It uses this heuristic:
# No upgrade work is necessary during a fresh system installation.
# But since linux-base is a new dependency of linux-image-* and did
# not exist until needed for the libata transition, we cannot simply
# test whether this is a fresh installation of linux-base. Instead,
# we test:
# - does /etc/fstab exist yet (this won't even work without it), and
# - are any linux-image-* packages installed yet?
sub is_fresh_installation {
if (-f '/etc/fstab') {
for (`dpkg-query 2/dev/null --showformat '\${status}\\n' -W
'linux-image-*'`) { return 0 if / installed\n$/;
}
}
return 1;
}
If you think this gives the wrong answer, please propose an alternative.
I cant come up with a reliable one. As we already discussed, not
having /var/log/installer must not mean anything :/
Note that if you hack the lenny installer to install a backported kernel
and suppress the libata transition code in linux-base, this may result
in configuration files that use unstable device names. You should
consider using the testing installer instead.
That's not feasable due to other customisations we've done. The above
workaround works for us, so...
cheers,
Holger
signature.asc
Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
Hi, On Samstag, 3. Juli 2010, Bastian Blank wrote: Well, then its easy: don't create this useless entry. ok, I'm now using a post-base-installer hook in d-i to remove the cdrom entry from /etc/fstab and voila, an inserted CDROM still pops up in KDE, yay. But I still wonder why this linux-base postinst bothers about fstab entries for /media/cdrom and /media/floppy at all. Those will never work with UUIDs :) cheers, Holger signature.asc Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
Hi, and I can still not preseed that warning away, that no bootloader has been detected... (Which is bad, as the .32-bpo is installed as part of the base system when this has been upgraded by d-i, but before the bootloader installation...) cheers, Holger signature.asc Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Thu, Aug 26, 2010 at 06:45:30PM +0200, Holger Levsen wrote: Hi, On Samstag, 3. Juli 2010, Bastian Blank wrote: Well, then its easy: don't create this useless entry. ok, I'm now using a post-base-installer hook in d-i to remove the cdrom entry from /etc/fstab and voila, an inserted CDROM still pops up in KDE, yay. But I still wonder why this linux-base postinst bothers about fstab entries for /media/cdrom and /media/floppy at all. Those will never work with UUIDs :) It may complain about an ATAPI floppy, but not one handled by a platform floppy driver. As for /media/cdrom, its device should be specified as /dev/cdrom or similar. Ben. -- Ben Hutchings We get into the habit of living before acquiring the habit of thinking. - Albert Camus -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100826172647.go5...@decadent.org.uk
Bug#587887: linux-base: please suppress useless debconf messages on first install
Hi Ben, On Donnerstag, 26. August 2010, Ben Hutchings wrote: It may complain about an ATAPI floppy, but not one handled by a platform floppy driver. As for /media/cdrom, its device should be specified as /dev/cdrom or similar. with the lenny installer, /dev/scd0 shall be mounted at /media/cdrom0 (using a vmware scsi CD-ROM drive..) cheers, Holger signature.asc Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Thu, Aug 26, 2010 at 07:40:02PM +0200, Holger Levsen wrote: Hi Ben, On Donnerstag, 26. August 2010, Ben Hutchings wrote: It may complain about an ATAPI floppy, but not one handled by a platform floppy driver. As for /media/cdrom, its device should be specified as /dev/cdrom or similar. with the lenny installer, /dev/scd0 shall be mounted at /media/cdrom0 (using a vmware scsi CD-ROM drive..) And that needs to be fixed. Ben. -- Ben Hutchings We get into the habit of living before acquiring the habit of thinking. - Albert Camus -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100826175921.gq5...@decadent.org.uk
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Donnerstag, 26. August 2010, Ben Hutchings wrote: with the lenny installer, /dev/scd0 shall be mounted at /media/cdrom0 (using a vmware scsi CD-ROM drive..) And that needs to be fixed. does the fix work with the kernels in lenny? signature.asc Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Thu, 2010-08-26 at 20:07 +0200, Holger Levsen wrote: On Donnerstag, 26. August 2010, Ben Hutchings wrote: with the lenny installer, /dev/scd0 shall be mounted at /media/cdrom0 (using a vmware scsi CD-ROM drive..) And that needs to be fixed. does the fix work with the kernels in lenny? Yes, the kernel in lenny supports symlinks. Really you should be asking about udev, and that does create the symlinks /dev/cdrom etc. in lenny. Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Thu, 2010-08-26 at 18:59 +0200, Holger Levsen wrote:
Hi,
and I can still not preseed that warning away, that no bootloader has been
detected...
(Which is bad, as the .32-bpo is installed as part of the base system when
this has been upgraded by d-i, but before the bootloader installation...)
I think we've discussed this before. linux-base is supposed to be quiet
during a fresh system installation. It uses this heuristic:
# No upgrade work is necessary during a fresh system installation.
# But since linux-base is a new dependency of linux-image-* and did
# not exist until needed for the libata transition, we cannot simply
# test whether this is a fresh installation of linux-base. Instead,
# we test:
# - does /etc/fstab exist yet (this won't even work without it), and
# - are any linux-image-* packages installed yet?
sub is_fresh_installation {
if (-f '/etc/fstab') {
for (`dpkg-query 2/dev/null --showformat '\${status}\\n' -W
'linux-image-*'`) {
return 0 if / installed\n$/;
}
}
return 1;
}
If you think this gives the wrong answer, please propose an alternative.
Note that if you hack the lenny installer to install a backported kernel
and suppress the libata transition code in linux-base, this may result
in configuration files that use unstable device names. You should
consider using the testing installer instead.
Ben.
--
Ben Hutchings
Once a job is fouled up, anything done to improve it makes it worse.
signature.asc
Description: This is a digitally signed message part
Bug#587887: linux-base: please suppress useless debconf messages on first install
Hi Ben, On Freitag, 2. Juli 2010, Ben Hutchings wrote: [/var/log/installer] We can't use this test. Older installations don't have such a directory. I dont understand. Every lenny (at least) installation done with d-i has this directory?! How are older installations relevant? cheers, Holger signature.asc Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Sat, Jul 3, 2010 at 12:54:12 +0200, Holger Levsen wrote: How are older installations relevant? How are they not? Cheers, Julien signature.asc Description: Digital signature
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Sat, 2010-07-03 at 12:54 +0200, Holger Levsen wrote: Hi Ben, On Freitag, 2. Juli 2010, Ben Hutchings wrote: [/var/log/installer] We can't use this test. Older installations don't have such a directory. I dont understand. Every lenny (at least) installation done with d-i has this directory?! How are older installations relevant? What you're suggesting will result in suppressing the upgrade questions if the system was originally installed with some earlier Debian release. Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part
Bug#587887: linux-base: please suppress useless debconf messages on first install
Hi, On Samstag, 3. Juli 2010, Julien Cristau wrote: On Sat, Jul 3, 2010 at 12:54:12 +0200, Holger Levsen wrote: How are older installations relevant? How are they not? Well, first I'm fairly very sure that even etch also created /var/log/installer... in that sense I dont think it matters, if you automatically install sarge with a 2.6.32.bpo kernel and get those warnings... cause I doubt anyone does that, thus I think those are not relevant. cheers, Holger signature.asc Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Fri, Jul 02, 2010 at 11:12:19AM +0200, Holger Levsen wrote: h01ger hi. i'm using .32.bpo on lenny, together with linux-base. so far so good, works great. but i have an issue with automated installs (with d-i and preseeding): linux-base informs me that /etc/fstab contains an entry (/etc/scd0 iirc) which might not work in future versions. i couldnt care less and would like to get rid of this warning,as it prevents fully automated installations Well, then its easy: don't create this useless entry. The device is named /dev/cdrom or so. Bastian -- ... freedom ... is a worship word... It is our worship word too. -- Cloud William and Kirk, The Omega Glory, stardate unknown -- To UNSUBSCRIBE, email to debian-kernel-requ...@lists.debian.org with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/20100703134316.gb14...@wavehammer.waldi.eu.org
Bug#587887: linux-base: please suppress useless debconf messages on first install
package: linux-base
version: 2.6.32-15
severity: wishlist
tags: patch
Hi,
first of all: thank you all for maintaining linux-2.6 - you're doing an
awesome job!
h01ger hi. i'm using .32.bpo on lenny, together with linux-base. so far so
good, works great. but i have an issue with automated installs (with d-i and
preseeding): linux-base informs me that /etc/fstab contains an entry
(/etc/scd0 iirc) which might not work in future versions. i couldnt care less
and would like to get rid of this warning,as it prevents fully automated
installations
bwh^ h01ger: I put a heuristic in there to work out whether this is a fresh
installation or upgrade
bwh^ h01ger: Maybe you can suggest how to improve it
bwh^ (the problem is that linux-base did not exist in lenny, so we cannot
check whether the package itself is being upgraded)
bwh^ Of course you should also be able to preseed the answer
h01ger i tried, using what debconf-get-selections |grep linux-base gave me
after an installation, but that didnt work
bwh^ weird
# make linux-base install quietly
#
# Boot loader configuration check needed
linux-base linux-base/disk-id-manual-boot-loader error
# Configuration files still contain deprecated device names
linux-base linux-base/disk-id-manual error
# Apply configuration changes to disk device IDs?
linux-base linux-base/disk-id-convert-plan-no-relabel boolean true
# Opdater diskenheds-id'er i systemkonfigurationen?
linux-base linux-base/disk-id-convert-auto boolean true
# Apply configuration changes to disk device IDs?
linux-base linux-base/disk-id-convert-plan boolean true
h01ger bwh, i wonder if your test fails as there were already .26 packages
installed...
h01ger ha! if /var/log/installer doesnt exist, its a fresh install. bingo
see attached patch
h01ger bwh, do you want a wishlist bug about this?
bwh^ h01ger: Yes, file a wishlist bug, please
cheers,
Holger
1578a1579,1581
if !(-d '/var/log/installer') {
return 0;
}
signature.asc
Description: This is a digitally signed message part.
Bug#587887: linux-base: please suppress useless debconf messages on first install
On Fri, 2010-07-02 at 11:12 +0200, Holger Levsen wrote: [...] h01ger bwh, i wonder if your test fails as there were already .26 packages installed... h01ger ha! if /var/log/installer doesnt exist, its a fresh install. bingo see attached patch We can't use this test. Older installations don't have such a directory. Ben. -- Ben Hutchings Once a job is fouled up, anything done to improve it makes it worse. signature.asc Description: This is a digitally signed message part
