live-{server,desktop}: USE_KDUMP=0 only live layer

kdump-tools uses ucf for config file management and naively
modifying the config file meant for the target system directly
will cause the file hash to not get updated in the ucf database.
This will then cause later modifications to fail because
"there's nothing to do". Although actually doing the modification
to the ucf database is messy. Let's just modify the file in the live
layer to get the behavior we want there.
ubuntu/oracular
Chris Peterson 3 months ago
parent c16a2fda9b
commit ed1844eda1

@ -1,19 +1,27 @@
#!/bin/bash -ex #!/bin/bash -ex
# We install kdump-tools in the minimal layer but it's enabled by default. # We install kdump-tools in the minimal layer but it's enabled by default.
# We disable it here and let subiquity/curtin decide when to enable it later. # subiquity/curtin will later decide to either keep it enabled or disable it
# This is also important so that it's not in some semi-enabled state in the # in the target system, but let's ensure it's disabled in the live layer.
# live system where USE_KDUMP=1 but it's missing the crashkernel parameter on # We do this by manually modifying /etc/default/kdump-tools to use USE_KDUMP=0.
# the kernel command line. Let's be doubly sure and set USE_KDUMP=0. # Without this, kdump-tools is still technically disabled since it requires both:
# 1. crashkernel on the kernel command line
# 2. USE_KDUMP=1 in /etc/default/kdump-tools
# and the kernel command line in the live layer doesn't have the crashkernel
# parameter in the kernel command line, but this semi-disabled state isn't
# ideal. So let's be doubly sure and set USE_KDUMP=0.
#
# Note that kdump-tools relies on ucf for configuration file management. So
# while we could modify the value in the minimal layer, doing so in a way that
# doesn't make future dpkg-reconfigures hard is a huge hassle. You also have
# to make sure to do the same steps to every layer it may get included.
case ${PASS} in case ${PASS} in
ubuntu-server-minimal) ubuntu-server-minimal.ubuntu-server.installer.*)
;; ;;
*) *)
exit 0 exit 0
;; ;;
esac esac
# kdump-tools relies on USE_KDUMP=1 to be enabled, set to USE_KDUMP=0 for off.
sed -i 's/USE_KDUMP=1/USE_KDUMP=0/' /etc/default/kdump-tools sed -i 's/USE_KDUMP=1/USE_KDUMP=0/' /etc/default/kdump-tools

@ -3,19 +3,28 @@
set -eux set -eux
# We install kdump-tools in the minimal layer but it's enabled by default. # We install kdump-tools in the minimal layer but it's enabled by default.
# We disable it here and let subiquity/curtin decide when to enable it later. # subiquity/curtin will later decide to either keep it enabled or disable it
# This is also important so that it's not in some semi-enabled state in the # in the target system, but let's ensure it's disabled in the live layer.
# live system where USE_KDUMP=1 but it's missing the crashkernel parameter on # We do this by manually modifying /etc/default/kdump-tools to use USE_KDUMP=0.
# the kernel command line. Let's be doubly sure and set USE_KDUMP=0. # Without this, kdump-tools is still technically disabled since it requires both:
# 1. crashkernel on the kernel command line
# 2. USE_KDUMP=1 in /etc/default/kdump-tools
# and the kernel command line in the live layer doesn't have the crashkernel
# parameter in the kernel command line, but this semi-disabled state isn't
# ideal. So let's be doubly sure and set USE_KDUMP=0.
#
# Note that kdump-tools relies on ucf for configuration file management. So
# while we could modify the value in the minimal layer, doing so in a way that
# doesn't make future dpkg-reconfigures hard is a huge hassle. You also have
# to make sure to do the same steps to every layer it may get included.
case ${PASS:-} in case ${PASS:-} in
minimal) *.live)
;; ;;
*) *)
exit 0 exit 0
;; ;;
esac esac
# kdump-tools relies on USE_KDUMP=1 to be enabled, set to USE_KDUMP=0 for off.
sed -i 's/USE_KDUMP=1/USE_KDUMP=0/' /etc/default/kdump-tools sed -i 's/USE_KDUMP=1/USE_KDUMP=0/' /etc/default/kdump-tools

Loading…
Cancel
Save