rq-scheduler VS supervisor

RQ-Scheduler is another simpler alternative (rq/rq-scheduler: A lightweight library that adds job scheduling capabilities to RQ (Redis Queue) (github.com)) that appears to be good for such purposes. It's not immediately clear if it would suffer from the same issues, but it seems not (Redis manages issues with data loss well, a separate queue is used for the scheduled tasks, etc.). Is anyone aware of any drawbacks to using RQ-Scheduler for something like this?

Some thoughts: - Storing API keys in Redis with AOF and RDB persistence turned on is going to be way faster than storing those keys in Mongo. - Did you mean RQ (redis-queue)/django-rq? If so, it works well as long as you don't need a scheduler for cron-like tasks, which it doesn't include. You can add rq-scheduler for that though: https://github.com/rq/rq-scheduler - Make sure your redis instance has a password -- redis 6 supports ACLs as well - The problem with slow requests is that they tie up app server processes and usually also database connections. That may be fine with a small number of consumers, but if you point your web site at this API, you may run into problems. Consider that if an app server serving web site traffic is waiting for a slow request to your API, then both app servers are affected -- you're now holding resources on the web site and the API, effectively. - HTTP clients often use a default timeout value for requests, and it's a best practice to use such a timeout -- so you'll need to coach your partners consuming this API not to use timeouts for your API.

What I went with was having both a web server (Apache/Nginx) and PHP-FPM in the same container image, held together by Supervisor: http://supervisord.org/

In my case, the Dockerfile looks a bit like the following:

  # Whatever base web server image you want, Debian/Ubuntu based here

As you can see there are several methods of running devpi server including cron, launchd (OSX service), nginx, Windows service, and supervisord. It also has a systemd service file which we can use to manage the service easily as Ubuntu uses it for primary service management. First off though we're going to need a proxy script to ensure that devpi is running in the virtual environment:

If it's a linux box you can make it a systemctl service, or you could use http://supervisord.org/.

I used supervisord to start my server and the cloud SQL proxy within the same container. That should work for your use case too.

I convinced (previous) $dayjob to use it. It (nix) kind of hung around in the background with the team that used haskell for awhile, but became prime time when we needed to support a range of VMs running within client infrastructure that were in reality just running various python scripts under supervisord (http://supervisord.org/). The range of client machines (redhat, centos, debian, ubuntu all of different releases) with differing versions of python and supervisord were driving our support and devops teams crazy (but in a weird way - they thought they were being productive, and really enjoyed tweaking things to work with additional varieties of os...). Additionally, having to work around some minor pain points of supervisord (adding and removing config files and not interrupting running services) lead to the realisation that there was a perfectly good service manager at the bottom of the modern versions of these systems (systemd) and that nixos was just a nix wrapper around this systemd and it would only restart what actually changed...

Here's more information about the configuration file format, in case anyone is curious: http://supervisord.org/configuration.html

My problem is that it's not always immediately clear how software that would normally run as a systemd service could be launched in the foreground instead. It usually takes a bit of digging around.

This post is structured as follows: The first chapter Requirements and Design, describes the requirements for such an application, defines its processes, breaks it down into logical components, and proposes a data model. The second chapter Implementation, provides an introduction to key implementation issues, such as implementing asynchronous tasks and LDAP authentication. It also showcases the usage of Docker with Python SDK in the project, including network configuration, and describes the deployment configuration using supervisord. The final chapter summarizes the efforts and provides links to the code repositories.

I have set up supervisord running cron job. However based on the discussion in supervisord GitHub, it is not possible to redirect cron's spawned command output to supervisord.

The project MISP-Docker from Coolacid is providing a containerized version of the MISP solution. This all-in-one solution includes the frontend, background jobs, cronjobs and an HTTP Server (Nginx) all orchestrated by process manager tool called supervisor. External services such as the database and Redis aren’t part of the container but are necessary. We decided that this project is very a good starting point to scale the MISP on Kubernetes.