As a Blockchain QA Engineer, How Do You Deal With Flaky Smart-Contract Tests That Fail Only on CI?

BlockchainMentorYagiz

BlockchainMentorYagiz

@BlockchainMentor
Updated: Nov 13, 2025
Views: 219

I work as a blockchain QA engineer on a DeFi project, and lately my test runs have become completely unpredictable. They pass on my local machine, fail on CI, and sometimes pass again without any changes.

I’ve already checked the Solidity logic, but most issues seem tied to RPC node behavior, gas price changes, or simple network delays.

Even after adding retries and timeouts, a few tests still fail at random. I want our CI runs to feel stable instead of a gamble. For those who’ve tested smart contracts at scale, how do you handle flaky test failures that come from network conditions rather than actual code problems?

Like 2 Replies 3

Replies

Welcome, guest

Join ArtofBlockchain to reply, ask questions, and participate in conversations.

Sign up Log in

ArtofBlockchain powered by Jatra Community Platform

  • Merrythetechie

    Merrythetechie

    @Merrythetechie • Nov 3, 2025

    We faced this in a Layer-2 project. I marked network-sensitive tests separately so the CI wouldn’t mix them with core logic tests. Using tools like evm_increaseTime and Hardhat helpers reduced most time-based failures. Keeping tests deterministic made results steady.

  • RubenzkArchitect

    RubenzkArchitect

    @zkArchitect • Nov 4, 2025

    Flaky tests often mean we’re not checking enough things. Earlier, I just checked for “no revert.” Now I log contract states before and after and verify post-conditions. Once I added those checks, random test failures dropped sharply.

  • Web3WandererAva

    Web3WandererAva

    @Web3Wanderer • Nov 4, 2025

    My tests failed randomly because Mocha ran them in a different order each time. Setting a fixed seed and sorting the tests made runs stable. It’s a small fix but saves a lot of time.