Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.maint.boot > #75788

Bug#1108660: partman-auto commands can cause loop

Cross-posted to 2 groups.

Show all headers | View raw

Package: preseed
Version: 1.124

With https://deb.debian.org/debian/dists/testing/main/installer-amd64/current/images/netboot/debian-installer/amd64/initrd.gz at the time of this post, the following preseed.cfg seems to cause debconf to go into an infinite loop:

d-i mirror/country string Australia
d-i mirror/http/hostname string mirror.aarnet.edu.au
d-i mirror/http/mirror select mirror.aarnet.edu.au
d-i mirror/http/directory string /pub/debian
d-i mirror/http/directory select /pub/debian
d-i mirror/http/proxy select
d-i netcfg/get_hostname debian-vm
d-i netcfg/get_domain debian-vm
d-i netcfg/choose_interface select auto
d-i mirror/suite string testing
d-i passwd/root-login boolean false
d-i passwd/user-fullname string voltagex
d-i passwd/username string voltagex
d-i passwd/user-password password changeme
d-i passwd/user-password-again password changeme
d-i time/zone string AU/Sydney
d-i partman-auto/init_automatically_partition select /dev/sda
d-i partman-auto/select_disk /dev/sda
d-i partman-auto/method regular
d-i partman-auto/choose_recipe select atomic

Boot parameters were priority=critical DEBCONF_DEBUG=5 country=AU language=en locale=en_AU.UTF-8 keymap=us but it's tricky to get logs from the VM in this state...

(it would be good to have a kernel parameter that puts the debug on its own serial terminal outside of GNU screen)

Output is something like

Jul  2 13:50:33 debconf: --> GO
Jul  2 13:50:33 debconf: <-- 0 ok
Jul  2 13:50:33 debconf: --> GET partman-auto/init_automatically_partition
Jul  2 13:50:33 debconf: <-- 0 50some_device__________regular
Jul  2 13:50:33 debconf: --> METAGET partman/text/scsi_disk description
Jul  2 13:50:33 debconf: <-- 0 SCSI%s (%s,%s,%s) (%s)
Jul  2 13:50:33 debconf: --> METAGET partman-auto/select_disk choices-c
Jul  2 13:50:33 debconf: <-- 0 /var/lib/partman/devices/=dev=sda
Jul  2 13:50:33 debconf: --> GET partman-auto/select_disk
Jul  2 13:50:33 debconf: <-- 0 /var/lib/partman/devices/=dev=sda
Jul  2 13:50:33 debconf: --> SET partman-auto/select_disk /var/lib/partman/devices/=dev=sda
Jul  2 13:50:33 debconf: <-- 0 value set
Jul  2 13:50:33 debconf: --> SUBST partman-auto/select_disk CHOICES /var/lib/partman/devices/=dev=sda
Jul  2 13:50:33 debconf: Adding [CHOICES] -> [/var/lib/partman/devices/=dev=sda]
Jul  2 13:50:33 debconf: <-- 0
Jul  2 13:50:33 debconf: --> SUBST partman-auto/select_disk DESCRIPTIONS SCSI1 (0\,0\,0) (sda) - 136.4 GB Msft Virtual Disk
Jul  2 13:50:33 debconf: Adding [DESCRIPTIONS] -> [SCSI1 (0\,0\,0) (sda) - 136.4 GB Msft Virtual Disk]
Jul  2 13:50:33 debconf: <-- 0
Jul  2 13:50:33 debconf: --> INPUT critical partman-auto/select_disk
Jul  2 13:50:33 debconf: <-- 30 question skipped
Jul  2 13:50:33 debconf: --> GO

over and over until the system OOMs.

I think it's the 
d-i partman-auto/init_automatically_partition select /dev/sda
d-i partman-auto/select_disk /dev/sda

causing the problem but shouldn't it cause a crash, error or warning instead of a loop?

(I couldn't find any documentation of how to make partman always choose the first disk on my VM)

Thanks,
Adam

Back to linux.debian.maint.boot | Previous | Next | Find similar | Unroll thread

Thread

Bug#1108660: partman-auto commands can cause loop Adam <debian@voltagex.org> - 2025-07-02 16:10 +0200

csiph-web