[baktun14]

Why

POST /v1/tx answered 500 five times on 2026-10-21 for transactions whose gas simulation the chain refused at message validation:

Query failed with (6): rpc error: code = Unknown desc = invalid pubkey value with gas used: '1150': unknown request
Query failed with (6): rpc error: code = Unknown desc = invalid deposit with gas used: '1150': unknown request

The chain runs message validation before the ante handler and returns the error as is, so the log never says failed to execute message. tx-signer only answers 400 when it does (#4052), so these went out as 500. console-api relayed them as "Service temporarily unavailable" and both 5xx alerts fired.

Fixes CON-1343

What

• Gas estimation also raises MessageExecutionFailedError when the Simulate rejection's description opens with "invalid". The error handler already answers that error with 400 and the chain's message.
• The match is anchored to the start of the description. Ante refusals the node raises on its own (an unordered ttl beyond the node's last block time ends in : invalid request) and recovered chain panics (recovered: runtime error: invalid memory address ...) stay 500.
• Sequence mismatches, expired simulations and transport failures keep their current answers.
• console-api needs no change: /v1/tx already relays a tx-signer 400 for caller-supplied messages (fix(managed-wallet): return the signer's 400 or 403 to the /v1/tx caller instead of a 500 #4057), and messages console builds itself still fail as 500.

Testing

• Unit tests cover three validation refusals, plus an unordered ttl refusal, a recovered panic and a node-level query refusal that must stay unclassified. A functional test has the mocked node refuse Simulate with invalid deposit and checks for a 400 with nothing broadcast.
• npm test, lint and tsc pass in apps/tx-signer. Stryker kills 5 of 5 mutants on the changed lines.

Summary by CodeRabbit

• Bug Fixes
  • Improved transaction simulation error classification for invalid messages and requests.
  • Preserved detailed chain error messages when simulation rejects a transaction.
  • Ensured invalid transactions return HTTP 400 responses without being broadcast.
  • Continued handling expiration, excessive TTL, recovered panics, and other refusal cases appropriately.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.26%. Comparing base (05f0098) to head (7b52d43).
⚠️ Report is 6 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4544      +/-   ##
==========================================
+ Coverage   94.07%   94.26%   +0.19%     
==========================================
  Files        1773     1697      -76     
  Lines       50951    48785    -2166     
  Branches    12442    11994     -448     
==========================================
- Hits        47934    45989    -1945     
+ Misses       2707     2501     -206     
+ Partials      310      295      -15     

Flag	Coverage Δ		*Carryforward flag
api	96.03% <ø> (ø)		Carriedforward from d5fd3e0
deploy-web	92.91% <ø> (ø)		Carriedforward from d5fd3e0
log-collector	?
notifications	95.95% <ø> (ø)		Carriedforward from d5fd3e0
provider-console	81.68% <ø> (ø)		Carriedforward from d5fd3e0
provider-inventory	?
provider-proxy	92.21% <ø> (ø)		Carriedforward from d5fd3e0
tx-signer	93.43% <100.00%> (+0.01%)	⬆️

*This pull request uses carry forward flags. Click here to find out more.

Files with missing lines	Coverage Δ
...-client-factory/signing-stargate-client.factory.ts	96.38% <100.00%> (+0.04%)	⬆️

... and 76 files with indirect coverage changes

🚀 New features to boost your workflow:

• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[claude]

Claude Code Review

This repository is configured for manual code reviews. Comment @claude review for a one-time review, or @claude review always to subscribe this PR to a review on every future push.

_{Tip: disable this comment in your organization's Code Review settings.}

[pullfrog]

This run hit its 1h time limit while still working. Give runs longer with --timeout=2h (or --notimeout) in this repo's instructions, or raise the workflow's timeout input if it sets one, since that input wins. A smaller PR or a lower effort also finishes sooner.

Repo settings → · Flags →

  ^{｜ Rerun failed job ➔ ｜ View workflow run ｜ via Pullfrog ｜ Using GLM-5.3}

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

⚙️ Run configuration

• Configuration used: Repository: akash-network/console/.coderabbit.yaml
• Review profile: CHILL
• Plan: Essentials
• Run ID: 0cc6937a-2e5d-411a-958c-70f45b434d90

📥 Commits

Reviewing files that changed from the base of the PR and between 05f0098 and 7b52d43.

📒 Files selected for processing: 3

• apps/tx-signer/src/lib/signing-stargate-client-factory/signing-stargate-client.factory.spec.ts
• apps/tx-signer/src/lib/signing-stargate-client-factory/signing-stargate-client.factory.ts
• apps/tx-signer/test/functional/sign-and-broadcast-tx.spec.ts

+----------------------------------------------------------------------------------------------------------------------------------------------+
| Eliminate effects between unrelated things. Design components that are self-contained, independent, and have a single, well-defined purpose. |
+----------------------------------------------------------------------------------------------------------------------------------------------+
  \
   \     /)/)
        ( '')
        /  |
      o(___)

✨ Finishing Touches

📝 Generate docstrings

• Commit to this branch
• Create a new PR

🧪 Generate unit tests (beta)

• Commit to this branch
• Create a new PR

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

_{Comment @coderabbitai help to get the list of available commands.}

[pullfrog]

✅ No new issues found.

Reviewed changes

Re-reviewed the delta since the previously reviewed d5fd3e0 — the single new commit 7b52d43, which tightens the invalid-message match to descriptions that open with "invalid".

• Anchored INVALID_MESSAGE_RE to the start of the description — a Simulate rejection now classifies as MessageExecutionFailedError only when its desc opens with invalid (desc = invalid\b). The three incident wordings stay 400 (invalid deposit, invalid pubkey value, Invalid groups via the /i flag), while ante refusals that merely end in a generic : invalid request (an unordered TTL past the node's 10-minute cap, … has already passed) and recovered chain panics keep answering 500. This supersedes the earlier optional suggestion to bound UNORDERED_TX_TTL_MS in env.config.ts by fixing the misclassification class itself.
• Added negative tests for both exclusion cases — a new it.each asserts the unordered-TTL and recovered-panic refusals stay unclassified with the chain's wording preserved and a single simulate attempt (no retry); the node-level refusal test also gained a no-retry assertion.

The fixture wording was verified against cosmos-sdk v0.53.4 (x/auth/ante/sigverify.go): verifyUnorderedNonce raises Wrapf(ErrInvalidRequest, "unordered tx ttl exceeds %s", …), which renders detail-first and ends : invalid request, so the anchor is the right discriminator; the ErrTxDecode/ErrUnauthorized wraps that do open with "invalid" are unreachable for cosmjs-built proto txs. Full tx-signer battery green locally: 179 tests (160 unit + 19 functional), npx tsc --noEmit, lint and prettier clean.

  ^{｜ View workflow run ｜ Using GLM-5.3 ｜ 𝕏}