globalBlockedRaiseException(ClusterBlockLevel level) (the non-project-aware flavour of this) has some 200 callers mostly Transport actions. These seem to be the bulk of what it takes to integrate the project global cluster blocks into existing checks for blocks. I've been through half of them and have update some based on the following.

• If a TransportAction's corresponding RestAction has a @ServerlessScope(Scope.PUBLIC) annotation, then I'd make its block check use this project aware version. In most cases, the action is already project aware or at least has a ProjectResolver, in some cases, I've added that as well.
• Some of the Scope.INTERNAL ones, probably still need to consider project block, e.g. TransportDeleteDataStreamLifecycleAction, TransportForceMergeAction. I've updated some as well.
• There are also some TransportActions that don't necessarily have a RestAction, but to me they seem to need to be project-aware w.r.t. checking blocks e.g. some of the persistent task related stuff such as CompletionPersistentTaskAction. I haven't dived into those.

I think properly going through all of these, is a rather big effort since in many cases, the relavent code is not project-aware and it is not clear to me if it will be. So my plan is to go through the second half of these as well and do the obvious ones. And open a ticket to come back to this. I assume we'd need to communicate this as part of "how to make things project-aware" when teams are doing their chuck of MP-compatibility.

globalBlockedRaiseException(ClusterBlockLevel level) (the non-project-aware flavour of this) has some 200 callers mostly Transport actions.

I think you mean globalBlockedException(ClusterBlockLevel).

CompletionPersistentTaskAction

This one is tricky since persistent task can be both cluster and project scoped. I think we might need something like the follows

if (projectResolver.supportsMultipleProjects() == false) {
    return state.blocks().globalBlockedException(ClusterBlockLevel.METADATA_WRITE);
} else {
    final var projectId = projectResolver.getProjectId();
    if (projectId == null || ProjectId.DEFAULT.equals(projectId)) {
        return state.blocks().globalBlockedException(ClusterBlockLevel.METADATA_WRITE);
    } else {
        return state.blocks().globalBlockedException(projectId, ClusterBlockLevel.METADATA_WRITE);
    }
}

We can iterate on this with a separate PR.

And open a ticket to come back to this

I agree. We don't own every action that uses the blocks. It makes sense to raise the awareness for other teams.

globalBlockedRaiseException(ClusterBlockLevel level) (the non-project-aware flavour of this) has some 200 callers mostly Transport actions.

I think you mean globalBlockedException(ClusterBlockLevel).

CompletionPersistentTaskAction

This one is tricky since persistent task can be both cluster and project scoped. I think we might need something like the follows

if (projectResolver.supportsMultipleProjects() == false) {
    return state.blocks().globalBlockedException(ClusterBlockLevel.METADATA_WRITE);
} else {
    final var projectId = projectResolver.getProjectId();
    if (projectId == null || ProjectId.DEFAULT.equals(projectId)) {
        return state.blocks().globalBlockedException(ClusterBlockLevel.METADATA_WRITE);
    } else {
        return state.blocks().globalBlockedException(projectId, ClusterBlockLevel.METADATA_WRITE);
    }
}

We can iterate on this with a separate PR.

And open a ticket to come back to this

I agree. We don't own every action that uses the blocks. It makes sense to raise the awareness for other teams.

Nit:

This is similar to CompletionPersistentTaskAction in that persistent tasks can be either project or cluster scoped. I'd leave this to a different PR that focuses on persistent tasks. What do you think?