Community of Practice: Running Pulp in Production

[bmbouter]

I’d like to invite users to participate in a working group where problems and solution encountered when running Pulp in production are shared.

To make this successful we need user participation. Please post a problem or question related to running Pulp in production as it’s own posting in the community section and then link to it in this thread.

This thread hopefully will be an aggregating place of links to info about problems and solution users have with Pulp in production.

[bmbouter]

Here are a few already-happening discussions related to production users:

• Pulp has to be entirely shut down at upgrade time. The zero-downtime working group is dedicated to solving that.

• Performance and Scale testing of Content Serving. See the test plan here.

[bmbouter]

Today a user shared that in their production environment they raised their yum/dnf timeouts from 2 seconds to 10 seconds. At 2 seconds the occasional package from pulp_rpm was taking just a little to long to cause timeouts. Raising the timeout resolved the issue.

Also FYI, here’s a pre-release demo of an upcoming way to replicate content between Pulp servers in multiple geographies with one Pulp server in each geo. This is called pulp-replica and it will be compatible with 3.23 as the replica, but it can pull content from earlier versions as the main Pulp server. A big thanks to @dkliban for pioneering this work.

[davidd]

I’ve added a topic here about a problem we’re running into with orphan cleanup in production:

Orphan cleanup problem

Inspired by the thread about running Pulp in Production, I figured I’d share a problem that we’ve run into and that’s with cleaning up orphans. Say you’re running Pulp in production for quite some time and you have a repo that’s collected hundreds of versions. Eventually you want to clean up these versions and their content. Pulp provides an automated way to do this so you don’t have to do it for each individual repo and that’s via “retain_repo_versions”. The problem is that the retain_repo_ve…

[bmbouter]

Here’s a blog post about pulp_rpm capacity planning and performance analysis for content serving. It can help determine how large (and how expensive) a cluster of pulp-content processes you need to run for a given traffic volume.