Conversation
pendo324
requested review from
Shubhranshu153,
Swapnanil-Gupta and
sondavidb
pendo324
force-pushed
the
mkosi-squashed
branch
from
4b7ca15 to
922098e
Compare
| @@ -0,0 +1,305 @@ | |||
| From 4597a8c36a5cd043fa27acfe90297a80c9a59f59 Mon Sep 17 00:00:00 2001 | |||
There was a problem hiding this comment.
Is this Patch required because mkosi doesnt have support for Amazon linux?
If yes what is the maintenance overhead for the patch, in the sense do we need to change if mkosi gets updated or changes some pattern?
This is a bit concerning if this is the case.
There was a problem hiding this comment.
The patch is taken from a PR which got abandoned. I don't anticipate much maintenance burden in mkosi since Amazon Linux is essentially treated as Fedora by mkosi, with a few small changes.
There was a problem hiding this comment.
Can we add documentation about the reference for the patch in readme.
And probably also the caveats for fixes done for AL2023.
Lets say a new AL version comes out do we have to redo the changes, but probably not relevant as don't see Al2023 going out of life any time soon
There was a problem hiding this comment.
Wondering if we could have used fedora or debian as the base distribution and gotten rid of the AL patch? Did we choose AL for familiarity or does it allow for an even smaller image size? Since we specify the packages to install in base configuration anyway, would it have made much difference?
There was a problem hiding this comment.
can you check my understanding is AL here is for the kernel and user space is fedora. But i might be wrong here.
deps/mkosi/README.md
Outdated
|
|
||
| ### Custom Package Building | ||
|
|
||
| - **`al2023-build/`**: Directory containing infrastructure for building custom AL2023 packages |
There was a problem hiding this comment.
nit: folder rename to al2023-custom-package-build
deps/mkosi/build.sh
Outdated
| @@ -0,0 +1,204 @@ | |||
| #!/bin/bash | |||
There was a problem hiding this comment.
this script probably can be split into single responsibility parts.
pendo324
force-pushed
the
mkosi-squashed
branch
from
922098e to
ef89ea2
Compare
|
@Shubhranshu153 addressed comments in latest revision. Had to rebase to fix the commit name |
pendo324
force-pushed
the
mkosi-squashed
branch
2 times, most recently
from
bbd5e89 to
64445b3
Compare
| git config --global --add safe.directory "${GITHUB_WORKSPACE}" | ||
| - name: Build QEMU | ||
| run: | | ||
| ln -s "$HOME/.docker" /root/.docker |
There was a problem hiding this comment.
nit: is the symlinking redundant?
| X86_64_ARTIFACT=Fedora-Cloud-Base-Generic-42-1.1.x86_64-20260107003429.qcow2 | ||
| X86_64_512_DIGEST=b5a6cb3361f437bd74b8b44f96e96709de440c1b41094b8125ca203928df665a8cdd4d8e58d0a81198da0d5f9a158917336a4a8b14b8eb3f33cd24c6278d1510 | ||
|
|
||
| # Cosign binary for image signing |
There was a problem hiding this comment.
nit: this can possibly be in it's own conf file. I can do a follow up on this.
| @@ -0,0 +1,305 @@ | |||
| From 4597a8c36a5cd043fa27acfe90297a80c9a59f59 Mon Sep 17 00:00:00 2001 | |||
There was a problem hiding this comment.
Wondering if we could have used fedora or debian as the base distribution and gotten rid of the AL patch? Did we choose AL for familiarity or does it allow for an even smaller image size? Since we specify the packages to install in base configuration anyway, would it have made much difference?
Swapnanil-Gupta
force-pushed
the
mkosi-squashed
branch
from
a67c453 to
91c3f10
Compare
Swapnanil-Gupta
force-pushed
the
mkosi-squashed
branch
from
91c3f10 to
c012e86
Compare
Signed-off-by: Justin Alvarez <alvajus@amazon.com>
Signed-off-by: Justin Alvarez <alvajus@amazon.com>
Signed-off-by: Justin Alvarez <alvajus@amazon.com>
Signed-off-by: Justin Alvarez <alvajus@amazon.com>
Signed-off-by: Justin Alvarez <alvajus@amazon.com>
Signed-off-by: Justin Alvarez <alvajus@amazon.com>
- no need for binfmt submodule update in build-qemu.sh because we already have a separate workflow for that - submodulesync.yaml - remove os updates from update-dependencies workflow as we have a new build-os workflow which will do the same - do not create os image update pr's if build-os workflow is running from a pr trigger - fix incorrent path in binfmt submodule update Signed-off-by: Swapnanil Gupta <swpnlg@amazon.com>
Swapnanil-Gupta
force-pushed
the
mkosi-squashed
branch
from
c012e86 to
afb5a38
Compare
|
LGTM. |
Signed-off-by: Swapnanil Gupta <swpnlg@amazon.com>
Swapnanil-Gupta
force-pushed
the
mkosi-squashed
branch
from
85e1281 to
d37435e
Compare
Signed-off-by: Swapnanil Gupta <swpnlg@amazon.com>
Signed-off-by: Swapnanil Gupta <swpnlg@amazon.com>
Signed-off-by: Swapnanil Gupta <swpnlg@amazon.com>
3 participants
Description of changes
Adds scripts, configuration, and a GitHub workflow for building Finch OS images using mkosi.
Testing done
Created working images for macOS and Windows using the GitHub Workflow: https://github.com/pendo324/finch-core/actions/runs/19875268071?pr=1
TODO
These are items which need to be done, but have been split off from this PR because it has already gotten too large:
Integrate these images with our existing tooling in finch/finch-core (e.g. make sure these images are tested)
Set up a schedule / triggers for automatically rebuilding the OS image
Integrate the dependencies in these images with existing version control / tracking mechanisms
I've reviewed the guidance in CONTRIBUTING.md
License Acceptance
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.