Skip to content

Retry the node health probe in the reusable deploy workflow - #5

Closed
nikolap-dat wants to merge 1 commit into
mainfrom
ci/health-check-retry
Closed

nikolap-dat wants to merge 1 commit into
mainfrom
ci/health-check-retry

Conversation

@nikolap-dat

@nikolap-dat nikolap-dat commented Oct 8, 2026 •

Copy link
Copy Markdown

Summary

arch-kit health samples the block height twice, 2 s apart, and fails when it did not move. A healthy chain still has multi-second gaps between blocks: on internal-b at v0.12.0, heights 48004 → 48005 were 3.5 s apart while the chain averaged ~9 blocks/s. prop-amm's deploy run 37778455694 landed its single probe in exactly that gap and failed with blocks are not progressing … height changed from 48005 to 48005 over 2s.

This makes the reusable workflow's health step probe up to five times, 10 s apart, failing only when every probe sees a stalled chain. The per-repo adapters (btc-wrapper, ausd-bridge, CLAMM, arch-vaults, arch-lend) already loop this way; prop-amm is the one consumer of the reusable workflow and will bump its pin to this commit.

Test plan

  • Workflow YAML parses (yq on the file)
  • prop-amm pin bumped to this merge commit and a real internal-b deploy passes the health step

🤖 Generated with Claude Code


View with [code]smith Autofix with [code]smith
Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.

`arch-kit health` samples the block height twice, 2 s apart, and fails when
it did not move. A healthy chain still has multi-second gaps between blocks
(3.5 s between heights 48004 and 48005 on internal-b at v0.12.0), so a single
probe fails the whole deploy whenever it lands in one. prop-amm's deploy run
37778455694 died this way on a chain producing ~9 blocks/s.

Probe up to five times, 10 s apart, and fail only when every probe sees a
stalled chain. The per-repo adapters already do this; this brings the
reusable workflow in line with them.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@nikolap-dat

Copy link
Copy Markdown
Author

Closing: the single-probe flake is accepted; prop-amm re-dispatches on failure.

@nikolap-dat nikolap-dat closed this Oct 8, 2026
@nikolap-dat
nikolap-dat deleted the ci/health-check-retry branch October 8, 2026 12:53
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