Skip to content

nodepath: move the node state root to /var/lib/ate - #1926

Merged
Benjamin Elder (BenTheElder) merged 1 commit into
agent-substrate:mainfrom
dims:pr/node-state-root
Sep 28, 2026
Merged

Benjamin Elder (BenTheElder) merged 1 commit into
agent-substrate:mainfrom
dims:pr/node-state-root

Conversation

@dims

@dims Davanum Srinivas (dims) commented Sep 27, 2026 •

Copy link
Copy Markdown
Collaborator

Fixes #1911

/var/lib/ateom-gvisor was named when gVisor was the only sandbox class; worker pods of both classes mount it. This renames nodepath.BasePath to /var/lib/ate everywhere it is spelled out: atelet's manifest, the kind CSI scripts, ate-setup's CSI step, and the docs. The controller's worker pod mounts follow the constant.

No compatibility path, per the comment above. A rolling upgrade rolls the pools once at the controller step, and a worker that lands on a node whose atelet still uses the old path reaches it only once that node moves; docs/upgrade.md says so. The old directory can be deleted afterwards.

  • Tests pass
  • Appropriate changes to documentation are included in the PR

@dims

Copy link
Copy Markdown
Collaborator Author

Do we want to do this? Tim Hockin (@thockin) Benjamin Elder (@BenTheElder)

@thockin

Copy link
Copy Markdown
Collaborator

I am AFK most of today, but we don't have to be so careful with compat (yet).

/var/lib/ateom-gvisor predates the micro-VM class; worker pods of both
classes mount it. Every component, manifest, script and doc moves to the
new name. No compatibility path: a rolling upgrade rolls the pools once
at the controller step, and a worker on a node whose atelet still uses
the old path reaches it only once that node moves.
@dims

Copy link
Copy Markdown
Collaborator Author

but we don't have to be so careful with compat (yet).

thanks Tim Hockin (@thockin) reworked it

@dims
Davanum Srinivas (dims) marked this pull request as ready for review September 27, 2026 18:26

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

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

We might put something in the docs about how to bulk cleanup existing installs but generally I agree we don't need backcompat at this point.

@BenTheElder
Benjamin Elder (BenTheElder) added this pull request to the merge queue Sep 28, 2026
Merged via the queue into agent-substrate:main with commit c7dbe9d Sep 28, 2026
11 checks passed
@bowei Bowei Du (bowei) added area/dev-infra area/node kind/cleanup Small fixes that are not bugs, for example a typo in a code comment labels Sep 29, 2026
kubernetes-prow Bot pushed a commit to kubernetes-sigs/gcp-filestore-csi-driver that referenced this pull request Oct 29, 2026
Agent Substrate moved its node state root from /var/lib/ateom-gvisor to
/var/lib/ate (agent-substrate/substrate#1926) and now asks the CSI node
plugin to mount actor volumes at /var/lib/ate/actors/<uid>/volumes/<name>.
The overlay only propagated /var/lib/ateom-gvisor, so on current Substrate
the NFS mount stayed inside the node plugin container and the actor saw an
empty directory.

Keep /var/lib/ateom-gvisor for Substrate release-0.1 and add /var/lib/ate
with the same bidirectional propagation.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area/dev-infra area/node kind/cleanup Small fixes that are not bugs, for example a typo in a code comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Rename the node state root /var/lib/ateom-gvisor to /var/lib/ate

4 participants