This repository was archived by the owner on Aug 3, 2026. It is now read-only.
Fix HOSB execution-sandbox end-to-end + rejected-queue GC - #80
Merged
Conversation
- Dockerfile.sandbox: COPY miner/ (validator/__init__ imports miner.submit) - container_argv: drop dup ENTRYPOINT prefix in both sandbox callers - _hosb_sandbox_nlls: chmod grid/out tempdirs for the userns-remap container - build_container_env: TMPDIR=/scratch (tmpfs root; /scratch/tmp absent) - archive_bundle: GC rejected checkpoints (RALPH_QUEUE_GC_REJECTED), keep manifests Verified e2e on the live king bundle: val_bpb 1.2112, benchmark 0.975. Co-Authored-By: Claude Opus 4.8 <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 subscribe to this conversation on GitHub.
Already have an account?
Sign in.
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
Makes the HOSB execution-sandbox (op4
RALPH_HOSB=shadow/enforce) run end-to-end, and stopsqueue/rejected/from filling the disk.The sandbox path had never run e2e; under the required
userns-remapit failed at every step. Each fix was found bringing it up on the live validator box.Fixes
COPY miner/—validator/__init__importsminer.submit, so the build-time import verify failed._hosb_sandbox_nlls+_sandboxed_hidden_eval): drop the duplicatedpython -m validator.sandbox_evalprefix; the imageENTRYPOINTalready runs it, so args were doubled andargv[1]became"python"(rc=3).chmodthe grid/outmkdtempdirs (0700root) so the remapped non-root container can read/grid+ write/out./scratch(the tmpfs root);/scratch/tmpis never created → "no usable temporary directory" (rc=2).archive_bundleremoves a rejected bundle'straining/checkpoint on archive (terminal + recorded inhf_state; manifests kept).RALPH_QUEUE_GC_REJECTED=0to opt out. The queue had grown to 51G (100% full).Verification
End-to-end on the live king bundle through the hardened container: val_bpb 1.2112, benchmark 0.975 (30s). Validator now runs
RALPH_HOSB=shadowcleanly.🤖 Generated with Claude Code