fix: the snapshot follows the run; the GPU probe reads total_memory - #206
Merged
Merged
Conversation
Two of the standby items. The session snapshot (conversation.json) was written only after a conversation turn, so a run driven from the Operate pane with nobody talking to the agent left it where the last chat ended — on the rig, at t1 while the run was at t15 — and nothing wrote it when the process went away: Ctrl-C, the desktop shell's quit and SIGTERM all reach the same finally block, and it stopped the server without saving. Now a landing volume saves it, off the loop and at most once a minute (it is one JSON write of the whole conversation), and shutdown saves it before the server stops. The mesh GPU probe read props.total_mem; torch's attribute is total_memory. The AttributeError fell into the blanket except, so the node never listed its GPU. This was also the one standing mypy error. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two of the standby items.
The snapshot follows the run.
conversation.jsonwas written only after a conversation turn, so a run driven from the Operate pane with nobody talking to the agent left it where the last chat ended (on the rig: embryo counts at t1 while the run was at t15), and nothing wrote it when the process went away: Ctrl-C, the desktop shell's quit and SIGTERM all reach the samefinallyblock, which stopped the server without saving. Now a landing volume saves it, off the event loop and at most once a minute (it is one JSON write of the whole conversation), and shutdown saves it before the server stops. The checkpoint (timelapse.yaml) still wins on resume, as #203 made it; this just stops the snapshot lying.The GPU probe.
props.total_mem→props.total_memory. The old name raisedAttributeErrorinto the blanketexcept, so the node never listed its GPU. This was also the one standing mypy error, souv run mypy .is now clean.Five tests in
tests/test_snapshot_follows_the_run.py.🤖 Generated with Claude Code