Skip to content

feat(dpuagent): write done marker file after all operations complete#53

Open
tsorya wants to merge 1 commit into
NVIDIA:public-mainfrom
tsorya:igal/dpu-agent-done-marker
Open

feat(dpuagent): write done marker file after all operations complete#53
tsorya wants to merge 1 commit into
NVIDIA:public-mainfrom
tsorya:igal/dpu-agent-done-marker

Conversation

@tsorya
Copy link
Copy Markdown

@tsorya tsorya commented May 24, 2026

Summary

  • The dpu-agent now writes a marker file at /run/dpu-agent/configuration-complete once all operations have completed successfully.
  • In OCP, we need a systemd dependency between the dpu-agent and kubelet so that kubelet only starts after the dpu-agent finishes all its configuration. This marker file enables other systemd units to use ConditionPathExists= or PathExists= to cleanly gate their startup on dpu-agent completion.
  • The marker is written to /run (tmpfs), so it is automatically cleared on reboot. It is written before the final status update to avoid blocking on a potentially long-running status push.

Test plan

  • Unit test: marker file is written after all operations complete successfully
  • Unit test: marker file is not written when the run is aborted (context cancellation)

In OCP, we need to create a systemd dependency between the dpu-agent
and kubelet so that kubelet only starts after the dpu-agent finishes
all its configuration tasks. To enable this cleanly in systemd, the
dpu-agent now writes a marker file at /run/dpu-agent/configuration-complete
once all operations have completed successfully. Other systemd units
can use a PathExists= or ConditionPathExists= directive on this file
to gate their startup on the dpu-agent being done.

The marker is written to /run (tmpfs), so it is automatically cleared
on reboot. It is written before the final status update to avoid
blocking on a potentially long-running status push.

Co-authored-by: Cursor <cursoragent@cursor.com>
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.

1 participant