Skip to content

fix(validator): restore the runner's Prometheus (bittensor_prometheus 1.0.3) - #858

Merged
kacper-wolkiewicz-reef merged 2 commits into
masterfrom
fix-validator-prometheus-entrypoint
Oct 26, 2026
Merged

kacper-wolkiewicz-reef merged 2 commits into
masterfrom
fix-validator-prometheus-entrypoint

Conversation

@kacper-wolkiewicz-reef

@kacper-wolkiewicz-reef kacper-wolkiewicz-reef commented Oct 26, 2026 •

Copy link
Copy Markdown
Contributor

Problem

Since the validator runner pinned backenddevelopersltd/bittensor_prometheus:1.0.2 (4785735, deployed to prod today), the runner's prometheus container restart-loops:

/bin/sh: 23: exec: /entrypoint.sh: Permission denied

The 1.0.2 image ships /entrypoint.sh as -rw-rw-r--, so validators that update send no metrics at all. The script was never executable in git, and the Dockerfile only ran chown on it. Fixed in bittensor-church/bittensor-prometheus-proxy#11 and released as 1.0.3.

Changes

  • bittensor_prometheus:1.0.2 → 1.0.3@sha256:35112f61…. latest on Docker Hub now points at the same build, which also heals validators still on the old runner compose.
  • bittensor-prometheus-proxy:1.0.2 → 1.0.3@sha256:39caeb31…. This is the same release, built by CI from tag v1.0.3; the proxy code is unchanged.
  • exec /entrypoint.sh … → exec sh /entrypoint.sh …, so the runner no longer depends on the script's file mode.

Testing

  • bittensor_prometheus with a throwaway wallet and a minimal config template, using the runner's exec /entrypoint.sh … call:
    • 1.0.2: Permission denied,
    • 1.0.3: /entrypoint.sh is -rwxr-xr-x, Prometheus logs "Server is ready to receive web requests.", and the hotkey is substituted into prometheus.yml.
  • exec sh /entrypoint.sh also starts Prometheus with the old 1.0.2 image.
  • Both pinned 1.0.3 digests pull from their registries. The proxy image imports pylon_client and bittensor.
  • The runner compose still passes docker compose config.

🤖 Generated with Claude Code

bittensor_prometheus:1.0.2 ships /entrypoint.sh without the executable bit,
so the Prometheus container restart-loops with "exec: /entrypoint.sh:
Permission denied" and validators stop reporting metrics. Running the script
through sh works with the pinned image and keeps working with future builds.
bittensor_prometheus 1.0.3 ships an executable /entrypoint.sh, fixing the
"Permission denied" restart loop of 1.0.2. Bump bittensor-prometheus-proxy
to the same release.
@kacper-wolkiewicz-reef kacper-wolkiewicz-reef changed the title fix(validator): start the runner's Prometheus through sh fix(validator): restore the runner's Prometheus (bittensor_prometheus 1.0.3) Oct 26, 2026
@github-actions

Copy link
Copy Markdown

Commit message lint report

gitlint --commits 47f00701a9499884cd8fe59a24bef164a993f965..c367fe00b16b3634a10d2c73b7c0b8f032a60ded

Status: ❌ Issues found.

gitlint output:

Linting commit range: 47f00701a9499884cd8fe59a24bef164a993f965..c367fe00b16b3634a10d2c73b7c0b8f032a60ded
gitlint --commits "47f00701a9499884cd8fe59a24bef164a993f965..c367fe00b16b3634a10d2c73b7c0b8f032a60ded"
Commit c367fe00b1:
1: UR1 Commit message is missing an 'Impacts:' footer. Add a line like: Impacts: validator, miner

Commit dc72c07b1c:
1: UR1 Commit message is missing an 'Impacts:' footer. Add a line like: Impacts: validator, miner
Gitlint found issues in the commit messages.

Fix by amending/rewording your commits (e.g., using git rebase -i and git commit --amend).

@kacper-wolkiewicz-reef
kacper-wolkiewicz-reef merged commit d4d01df into master Oct 26, 2026
31 of 32 checks passed
@kacper-wolkiewicz-reef
kacper-wolkiewicz-reef deleted the fix-validator-prometheus-entrypoint branch October 26, 2026 16:19
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