Skip to content

Dev fixes#28

Merged
mmatthiesencsc merged 28 commits into
mainfrom
dev-fixes
Apr 1, 2026
Merged

Dev fixes#28
mmatthiesencsc merged 28 commits into
mainfrom
dev-fixes

Conversation

@Traubert
Copy link
Copy Markdown
Contributor

@Traubert Traubert commented Jan 8, 2026

Various issues prevented the client commands from completing successfully. This branch at least makes things complete, though some things should still be fixed more correctly.

Many changes have stupidly been piled on this same PR, we'll review with some team and see if this is mergeable.

- Add get_vault_client() helper to refresh JWT tokens before Vault operations
  (JWT SVIDs expire after 1h, Vault tokens after 24h)
- Change workload selector from docker:image_id: to unix:uid:0
  (Docker attestor not working, all client containers run as root)
- Add chmod 777 for SPIRE agent socket after creation
  (Fixes socket permission issues preventing SVID fetching)

These changes fix the client registration failures and allow data/container/job
preparation workflows to complete successfully.
Container building can somewhat confusingly fail when the sif image
already exists, or when the host /tmp ends up getting mounted to the
build container's /output. These are intended to defaultly avoid those
problems, and provide some configurability.
Traubert added 17 commits March 25, 2026 15:15
Previously, "follow" meant to stream job log output, and otherwise we
persisted with the job, printing changes in its status. Now, "follow"
means the old default state, --verbose means to stream the job log,
and the default is to exit after the job has been submitted.
First wait, then check if process died, then print that we're waiting
some more. This way there's usually no concerning needless printing.
(Because of the need to use our own binary which may not be able to
access system's LDAP, this is why we also need to use UID attestation
instead of username)
@Traubert Traubert changed the title WIP dev fixes Dev fixes Mar 27, 2026
Copy link
Copy Markdown
Collaborator

@mmatthiesencsc mmatthiesencsc left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good, great steps ahead. I created #29 but that is for later to decide.

Comment thread server/app.py Outdated
Comment thread client/container_preparation/entrypoint.sh
@mmatthiesencsc
Copy link
Copy Markdown
Collaborator

The checks fail in part because of missing LFS budget. Ignoring for now.

@mmatthiesencsc mmatthiesencsc merged commit 32a3e96 into main Apr 1, 2026
2 of 8 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants