fix(cluster): gg dns records leak if server destroyed before install complete

Author: NathanFlurry

svc/pkg/cluster/worker/src/workers/server_destroy.rs

@@ -70,21 +70,6 @@ async fn inner(
 		server.pool_type as i32
 	));
 	if let backend::cluster::PoolType::Gg = pool_type {
-		// Update db record
-		sql_execute!(
-			[ctx]
-			"
-			UPDATE db_cluster.servers_cloudflare
-			SET destroy_ts = $2
-			WHERE
-				server_id = $1 AND
-				destroy_ts IS NULL
-			",
-			server_id,
-			util::timestamp::now(),
-		)
-		.await?;
-
 		msg!([ctx] cluster::msg::server_dns_delete(server_id) {
 			server_id: ctx.server_id,
 		})

svc/pkg/cluster/worker/src/workers/server_dns_delete.rs

@@ -63,10 +63,10 @@ async fn inner(
 		secondary_dns_record_id,
 	}) = row
 	else {
+		// TODO: This might get stuck in a loop if does not finish installing
 		// NOTE: It is safe to do nothing in this case because both this worker and
 		// `cluster-server-dns-create` use transactions
-		tracing::warn!("server has no dns records");
-		return Ok(());
+		retry_bail!("server has no dns records yet");
 	};
 
 	let zone_id = unwrap!(util::env::cloudflare::zone::job::id(), "dns not configured");