anaconda: bios boot for container-installer #360
Conversation
supakeen
force-pushed
the
iso-bios-boot
branch
from
ae52897
to
48770d5
Compare
|
This PR enables BIOS installation without "kickstart insufficient". |
| @@ -390,6 +390,8 @@ keyboard us | |||
| timezone UTC | |||
|
|
|||
| clearpart --all | |||
|
|
|||
| part biosboot --size=1 --fstype=biosboot | |||
There was a problem hiding this comment.
I believe we should be able to use reqpart too now.
| @@ -390,6 +390,8 @@ keyboard us | |||
| timezone UTC | |||
|
|
|||
| clearpart --all | |||
|
|
|||
| part biosboot --size=1 --fstype=biosboot | |||
| part /boot/efi --fstype=efi --size=512 --fsoptions="umask=0077" | |||
| part /boot --fstype=ext2 --size=1024 --label=boot | |||
There was a problem hiding this comment.
We really really should not use ext2 here...there's no filesystem journaling. Is this a hack-around for osbuild/bootc-image-builder#20 ?
If so, then...we need a reminder to remove it ASAP. If it's not, please fix to use ext4.
Of course this larger topic goes really into the rampant duplication we have of the filesystem layout in kickstarts, many times in Go code in this repo, anaconda, and bootc, and fedora-coreos etc.
What I think we should aim for is moving the default partitioning into the container - not hardcoded in this repo or in bootc code. Doing that with anaconda would require rhinstaller/anaconda#5197
There was a problem hiding this comment.
Agreed it feels icky, it was taken directly from a pre-existing kickstart which only solidifies the problem. Tracking here: #362.
No description provided.