Hi everyone,
I’m running into some frustrating stability issues with our current CI pipeline and was wondering if anyone else has optimized their resource classes for similar workloads. We’re primarily using CircleCI to automate our testing and deployment cycles for https://capcapk.com, but we’ve hit a wall with Out-of-Memory (OOM) errors during the apk assembly phase.
The issue seems to trigger specifically when the pipeline attempts to run parallelized UI tests on the Android emulator. Even with the medium+ resource class, the gradlew process frequently spikes and kills the job before the artifacts can be uploaded. I’ve tried toggling the --no-daemon flag and limiting the Java heap size with _JAVA_OPTIONS: "-Xmx2048m", but the build still feels incredibly fragile.
Has anyone found a “sweet spot” for configuring the android orb to handle large asset libraries without constantly jumping to the most expensive resource tiers? I’m also seeing a related bug where the cache isn’t being restored correctly after a failed OOM build, which forces a full clean build every time and really blows out our credit usage. If you’ve managed to stabilize resource-heavy Android builds or have a specific config.yml snippet for better memory management in these scenarios, I’d really appreciate the help!