Allow to pass in extra flags into the call to kexec

[johbo]

Noticed that the call to kexec was hanging in some cases when I had a machine which did have a stale rbd volume mounted.

Based on the logs I found out that the calls to kexec were stuck because of the stale volume mounts. Using the --no-sync flag does work fine for this case.

I am not 100% sure if there could be a good reason why one wants to avoid this flag in the context of booting into the installer. If there are reasons, then we could maybe allow to pass in parameters via the call to nixos-anywhere. Please let me know if this would be better, then i will tweak the change.

There is a related PR into nixos-anywhere:

• Add new parameter "--kexec-extra-flags" nixos-anywhere#332

[Mic92]

Also most people probably use this to replace the old system,
but in case someone just needs this to perform some filesystem maintenance,
we shouldn't uncleanly unmount the filesystem.

I thought about having a timeout:

Suggested change

	nohup sh -c "sleep 6 && '$SCRIPT_DIR/kexec' -e --no-sync" &
	nohup sh -c "sleep 6 && (timeout 30 sync; '$SCRIPT_DIR/kexec' -e --no-sync)" &

However this also is not ideal because unlike the kexec version this actually performs the sync before all userspaces processes are stopped.

So I think this should be a flag instead.

[Mic92]

I think we could have a way to allow to pass flags to kexec in general instead of just having a --no-sync flag.

[johbo]

I actually also think now that this would be better. Allowing to pass in flags makes it flexible for all sorts of corner cases. Will try to work this out and push an update into this PR.

[johbo]

@Mic92 hope I got it figured out, struggled a bit with passing the extra flags around in a proper way. Did introduce the sh -c so that it also works if multiple flags are specified.

[Mic92]

@mergify queue

[mergify]

queue

🛑 The pull request has been removed from the queue default

The pull request can't be updated.

You can take a look at Queue: Embarked in merge queue check runs for more details.

In case of a failure due to a flaky test, you should first retrigger the CI.
Then, re-embark the pull request into the merge queue by posting the comment
@mergifyio refresh on the pull request.

[Mic92]

Merged in #236