Gitlab database noticeable size increase since 16.3.3

How to Use GitLab Upgrade
1

Hi There,

I am self-hosting Gitlab EE Ultimate for personal projects, homelab, etc.
On the 18th of September, Gitlab EE 16.3.3 was installed by my automation. Because this is done automatically, I did not look much into this upgrade and Gitlab worked fine afterwards.
However, what I did not notice is that the Gitlab database was constantly increasing until it filled up the whole disk. Postgresql database size went from around 2 GB to 4.8 GB.
I have now upgraded to Gitlab ee 16.3.4. Database usage appears to have stopped.
I looked into the db itself, it might be due to pm_packages, pm_packages_versions and pm_packages_versions_licenses.
Would any of you noticed the same behavior? Is there anything I could do to reclaim this space? Or is it working as intended?
I have already tried to do a vacuum but it did not free up anything.

Kind regards,

2

Those tables have started to increase again.
Looking at the data, it seems that Gitlab selfhosted is pulling data from gitlab.com.
Here is a sample from pm_packages

id	purl_type	"name"	created_at	updated_at	licenses
2165253	4	github.com/tencentcloud/tencentcloud-sdk-go/tencentcloud/rum	2023-09-18 09:42:00.918 +1200	2023-09-21 17:38:18.422 +1200	[[2], "1.0.221", "1.0.752", []]
2167029	4	github.com/tencentcloud/tencentcloud-sdk-go/tencentcloud/eb	2023-09-18 09:42:10.084 +1200	2023-09-21 17:50:50.485 +1200	[[2], "1.0.326", "1.0.752", []]
2165211	4	github.com/kiegroup/kogito-operator/test	2023-09-18 09:41:59.877 +1200	2023-09-21 17:50:54.344 +1200	[[2], "0.0.0-20210916151339-fa5dd0381930", "0.0.0-20230919182244-28b2d3dc945e", []]
119539	1	designcafe/cafeapi	2023-09-18 05:41:43.245 +1200	2023-09-20 02:25:55.537 +1200	[[11], "1.0.0", "1.0.0", []]
119536	1	designbycode/guardian	2023-09-18 05:41:43.245 +1200	2023-09-20 02:25:55.537 +1200	[[11], "1.0.0", "1.4.1", []]

Any idea what is happening? I can give more storage to my instance but don’t see/understand the usefulness of this data.

3

Ok, last post from myself as I resolved this issue. Hoping this could help anybody else.
In 16.3.3, package metadata sync are enabled by default:
Backport Enable sync with package metadata db by default

To disable this:

  1. Go to Admin Area > Settings > Security and Compliance
  2. Expand License Compliance
  3. Untick all options (or keep those you need) and save changes
  4. You can then follow this documentation to delete downloaded packages: License scanning of CycloneDX files | GitLab
  5. Finally on postgresql server, you can run a vacuum or wait for auto vacuum

Signing off :slight_smile:

2 Likes