"Artifacts for pages are too large"

TL;DR: The artifacts do get uploaded, but apparently they fail to extract.

---

Docs are no longer deployed due to the pages:deploy failure "Artifacts for pages are too large".

• First pipeline that this occurred: https://gitlab.com/gitlab-org/gitlab-docs/pipelines/75057999
  • pages job https://gitlab.com/gitlab-org/gitlab-docs/-/jobs/267259506
  • artifacts zip size: 442MB
• Last successful pipeline: https://gitlab.com/gitlab-org/gitlab-docs/pipelines/75054338
  • pages job https://gitlab.com/gitlab-org/gitlab-docs/-/jobs/267245450
  • artifacts zip size: 439MB

As a comparison, the www-gitlab-com's artifact zip size is 767MB https://gitlab.com/gitlab-com/www-gitlab-com/-/jobs/267376429

The artifacts do get uploaded as opposed to gitlab-com/www-gitlab-com#4782 where they failed to upload in the first place. It's something to do with Pages I think. In www-gitlab-com we don't use Pages.

Questions

1. Does the artifacts size count the zipped file or its unzipped contents?
   • It's the uncompressed size that matters https://gitlab.com/gitlab-org/gitlab-ce/blob/006bc124366078d717ea8a331fdfd1e2ccd3476f/app/services/projects/update_pages_service.rb#L90

ASAP actions

The problem is that we have duplicate content for CE and EE, even more so now that the single codebase is a fact.

What we can do, is pull only from one repo and symlink ce/ to ee/. That would save almost 60MB per version, which is almost 250MB in total! That's half the size of what we have today.

Future

The following issues will help us in the future:

• Multiple version Pages support https://gitlab.com/gitlab-org/gitlab-ce/issues/35141
• GitLab Pages - Storage size Limitations by Project or Group https://gitlab.com/gitlab-org/gitlab-ce/issues/42962