You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our CI pipeline has an intermittent failure that occurs when building two dependent libraries in a integrated mono-repo. The following is the failure that occurs:
Here's the command being run (distributed across 5 nx build agents):
Strangely, this occurs only when building most/all of the projects in the mono-repo, for example, when the nx.json file has been modified. But this behaviour it not consistent - sometimes touching the nx.json results in a successful build of all projects. I suspect this is because the distribution of tasks sometimes result in the build of these dependent projects sometimes occurring on the same nx build agent, and sometimes not.
I have checked that the outputs fields in the project.json files for these two libraries accurately reflect the file locations generated by their build.
To be able to build dependent libraries, regardless of whether the build for each occurs on the same or different nx agent ie that the caching and replay of the library output should be consistent and reliable.
GitHub Repo
No response
Steps to Reproduce
Given the inconsistent nature of this problem, building a reproducible repo will involve many apps, each having e2e tests whose duration is between 3-10 minutes each, many libraries, with inter-dependencies. And still this may or may not reproduce the error.
Happy to provide access to the nx cloud logs (CIPE: 409256) as required. Or to try and reproduce the error again, with any additionally telemetry / logging switches enabled.
Current Behavior
Our CI pipeline has an intermittent failure that occurs when building two dependent libraries in a integrated mono-repo. The following is the failure that occurs:
Here's the command being run (distributed across 5 nx build agents):
Strangely, this occurs only when building most/all of the projects in the mono-repo, for example, when the nx.json file has been modified. But this behaviour it not consistent - sometimes touching the nx.json results in a successful build of all projects. I suspect this is because the distribution of tasks sometimes result in the build of these dependent projects sometimes occurring on the same nx build agent, and sometimes not.
I have checked that the
outputs
fields in the project.json files for these two libraries accurately reflect the file locations generated by their build.Summary for the two libraries
ask-agora-insights-angular
->ask-agora-insights
ask-agora-insights
: a senciljs library that:ask-agora-insights-angular
libraryask-agora-insights-angular
: an angular application that packages the generated component proxies fromask-agora-insights
ask-agora-insights
locally using remote cache, restores the output locations defined by the project.jsonThe project config for these two libraries
ask-agora-insights
project.json:
resolved project config:
ask-agora-insights-angular
project json:
resolved project config:
Expected Behavior
To be able to build dependent libraries, regardless of whether the build for each occurs on the same or different nx agent ie that the caching and replay of the library output should be consistent and reliable.
GitHub Repo
No response
Steps to Reproduce
Given the inconsistent nature of this problem, building a reproducible repo will involve many apps, each having e2e tests whose duration is between 3-10 minutes each, many libraries, with inter-dependencies. And still this may or may not reproduce the error.
Happy to provide access to the nx cloud logs (CIPE: 409256) as required. Or to try and reproduce the error again, with any additionally telemetry / logging switches enabled.
Nx Report
Failure Logs
Package Manager Version
No response
Operating System
Additional Information
No response
The text was updated successfully, but these errors were encountered: