Revisiting `reset_all` and `reset_all_tagged`

[cfzimmerman]

dpd-api currently marks both these with "TODO-security: This endpoint should probably not exist."

dendrite/dpd-api/src/lib.rs

Lines 1544 to 1573 in 2475028

	/// Clear all settings associated with a specific tag.
	///
	/// This removes:
	///
	/// - All ARP or NDP table entries.
	/// - All routes
	/// - All links on all switch ports
	// Note: This endpoint does not clear multicast groups.
	// TODO-security: This endpoint should probably not exist.
	#[endpoint {
	method = DELETE,
	path = "/all-settings/{tag}",
	}]
	async fn reset_all_tagged(
	rqctx: RequestContext<Self::Context>,
	path: Path<latest::misc::TagPath>,
	) -> Result<HttpResponseUpdatedNoContent, HttpError>;
	/// Clear all settings.
	///
	/// This removes all data entirely.
	// Note: Unlike `reset_all_tagged`, this endpoint does clear multicast groups.
	// TODO-security: This endpoint should probably not exist.
	#[endpoint {
	method = DELETE,
	path = "/all-settings"
	}]
	async fn reset_all(
	rqctx: RequestContext<Self::Context>,
	) -> Result<HttpResponseUpdatedNoContent, HttpError>;

I'm making adjacent changes to resource tagging, and I'd like to know whether to version-migrate the tagged endpoints or sunset them both. If the former, I'd like to replace the TODO comments. If the latter, I'll open a PR for removal.

tfportd is the only user I've found so far. Not clear to me how to migrate this if we decide to delete the endpoints. Open to suggestions.

dendrite/tfportd/src/main.rs

Lines 296 to 302 in 2475028

	// If the reset fails, we'll just log the error and continue, as we will
	// try to clean-up the any stale settings during the normal reconciliation
	// process. If those fail, we log an error and try again on a subsequent
	// iterations.
	if let Err(e) = client.reset_all_tagged(CLIENT_NAME).await {
	error!(log, "Error while flushing old state: {e:?}");
	}