Purge old Workspaces at some point

[geropl]

Description

Introduces a 3rd workspace deletion state "purged" as discussed in #11378 .

Related Issue(s)

Fixes #11378

How to test

Run tests

• (cd components/gitpod-db && yarn test)

Execute in preview environment

• open a workspace on this branch
• kubectl port-forward mysql-0 3306 &
• mysql -u gitpod -h 127.0.0.1 -p, PW: PW
• Insert these:

INSERT INTO `gitpod`.`d_b_workspace`
(`id`,
`creationTime`,
`ownerId`,
`contextURL`,
`description`,
`context`,
`config`,
`softDeleted`,
`softDeletedTime`,
`contentDeletedTime`,
`cloneURL`)
VALUES
('123',
'2019-09-25T21:44:13.560Z',
'123',
'https://example.org/my/repo',
'my-repo',
'{}',
'{}',
'gc',
'2019-09-25T21:44:13.560Z',
'2019-10-25T21:44:13.560Z',
'https://example.org/my/repo');

INSERT INTO `gitpod`.`d_b_workspace_instance`
(`id`,
`workspaceId`,
`region`,
`creationTime`,
`configuration`,
`ideUrl`,
`workspaceImage`,
`status`)
VALUES
('1231',
'123',
'dev',
'2019-09-25T21:44:13.560Z',
'{}',
'https://example.org/my/workspace',
'',
'{"phase": "stopped"}');

• kubectl edit cm server-config and set workspaceGarbageCollection.intervalSeconds to 30 (or similar)
• kubectl rollout restart deployment server
• wait at least 30 mins until you observe this line in the server logs: wsgc: successfully purged 1 workspaces

Release Notes

NONE

Documentation

Werft options:

• /werft with-preview

[werft-gitpod-dev-com]

started the job as gitpod-build-gpl-clean-wss.2 because the annotations in the pull request description changed
(with .werft/ from main)

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[svenefftinge]

looking into this, now 👀

[svenefftinge]

Looks really good! ❤️
Would be great to also delete corresponding records from

• d_b_prebuilt_workspace
• d_b_workspace_instance_user
• d_b_layout_data

Adding hold in case you want to address this in this PR.

/hold

[geropl]

Would be great to also delete corresponding records from

Yes indeed... make sense to do it now I guess. Have to double-check with @jakobhero about which ones are safe, though.

[geropl]

d_b_prebuild_info is another candidate as well.

[easyCZ]

@geropl This now has metrics also. I've bumped the retention to 3 years for now, to observe the effect.

[geropl]

@easyCZ Yes, looks good to me as well (cannot approve as it's my PR 🙈 )! 👍

Just to summarize:

• we decided to go with WorkspaceInstance and Workspace first
• d_b_prebuilt_workspace, d_b_workspace_instance_user, d_b_prebuild_info and d_b_layout_data are good candidates to add as well, but we exclude them for now to move forward. We can add them later, and handle the diff manually (e.g., identify entries without workspace, and schedule removal of those)

[easyCZ]

/unhold

💥