Redmine

WebhookJob runs on the default queue, like every other Redmine async job (except mails).

The issue : a single unreachable endpoint could quickly saturate this queue: each delivery waits for the 60second timeout before failing, so a worker with 3 threads processes at most 3 deliveries per minute. Meanwhile every issue update enqueues new ones, the queue grows faster than it drains, and all other jobs on the default queue wait behind them.

Email notifications would be spared only because the mailers queue has its own worker.

I propose to move webhook deliveries to their own queue, so that administrators can give them a dedicated worker and a failing endpoint cannot hold up the rest of the application:

--- a/app/jobs/webhook_job.rb
+++ b/app/jobs/webhook_job.rb
@@ -18,6 +18,8 @@
 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.

 class WebhookJob < ApplicationJob
+  queue_as :webhooks
+
   def perform(hook_id, payload_json)
     if hook = Webhook.find_by_id(hook_id)
       if hook.user&.active?

Administrators who list their queues explicitly already include the mailers queue (Action Mailer's default) ; they would also have to list the "webhooks" queue.