Skip to content

Regularly disable policies linked with no container repositories

David Fernandez requested to merge 330315-step-in-cron-worker into master

🌳 Context

In #330315 (closed), we described a ~performance problem where there were too many cleanup policies enabled for projects with no container images. This made some database indexes used by the background jobs that executes those policies less efficient.

In !61983 (merged), we created a background migration that bulk disable policies in those conditions.

Unfortunately, the background migration missed ~90 000 policies and we created a post deploy migration for those in !63841 (merged).

Now, it's time to disable those policies in a recurrent way. This way, we ensure that database indexes using those policies always reference the optimal set of policies (enabled policies with container repositories)

We have the perfect candidate for that: within the jobs dealing with cleanup policies, we have a cron job that runs each 50min. It can very well, lookup those policies and disable them.

🔍 What does this MR do?

  • Update the ContainerExpirationPolicyWorker (cron job) to disable those policies
  • Update the related specs
  • This process was documented in the breaking changes blog post for %14.0 but this recurrent process is now added to the documentation

📷 Screenshots (strongly suggested)

n / a

📐 Does this MR meet the acceptance criteria?

Conformity

Availability and Testing

Security

Does this MR contain changes to processing or storing of credentials or tokens, authorization and authentication methods or other items described in the security review guidelines? If not, then delete this Security section.

  • [-] Label as security and @ mention @gitlab-com/gl-security/appsec
  • [-] The MR includes necessary changes to maintain consistency between UI, API, email, or other methods
  • [-] Security reports checked/validated by a reviewer from the AppSec team

💾 Database Review

So basically, we need to update all expiration policies without any container repository. This update can be slow to be done and so, I used an each_batch loop to limit the number of rows that the query needs to go through.

Edited by David Fernandez

Merge request reports