Closed KitsuneAlex closed 1 year ago
ran into a panic when a job fails.
How does your job yaml look like? Just to help me to understand how to reproduce this scenario.
I don't know what happens in your case after a null check like in #152, it probably has more consequences for logging after this fault.
Thanks for the super quick reply! <3 My workflow for this particular instance looks as follows (i'm using a lot of reusable workflows):
name: Clang
on:
workflow_call:
inputs:
name:
required: true
type: string
platform_name:
required: true
type: string
std_version:
required: true
type: string
version:
required: true
type: string
cmake_args:
required: false
type: string
default: ""
ctest_args:
required: false
type: string
default: ""
build_type:
required: false
type: string
default: "Debug"
env:
BUILD_TYPE: ${{inputs.build_type}}
CC: "clang-${{inputs.version}}"
CXX: "clang++-${{inputs.version}}"
jobs:
build:
name: ${{inputs.name}}
runs-on: ["arm64"]
steps:
- uses: actions/checkout@v3
- name: Initialize Submodules
run: git submodule update --init
- name: Configure CMake (${{inputs.platform_name}})
run: >-
cmake -B ${{github.workspace}}/build-${{inputs.platform_name}}-clang-cpp${{inputs.std_version}}
-DCMAKE_CXX_FLAGS=-stdlib=libc++
-DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}}
-DGTEST_USE_OWN_TR1_TUPLE=1
-DCMAKE_CXX_STANDARD=${{inputs.std_version}}
${{inputs.cmake_args}}
- name: Build (${{inputs.platform_name}})
run: >-
cmake --build ${{github.workspace}}/build-${{inputs.platform_name}}-clang-cpp${{inputs.std_version}}
--config ${{env.BUILD_TYPE}}
--clean-first
- name: Test (${{inputs.platform_name}})
working-directory: ${{github.workspace}}/build-${{inputs.platform_name}}-clang-cpp${{inputs.std_version}}
run: ctest -C ${{env.BUILD_TYPE}} ${{inputs.ctest_args}}
Your workflow isn't causing anything odd here, sounds like a temporary logger inconsitency.
Your workflow isn't causing anything odd here, sounds like a temporary logger inconsitency.
Interesting. It happened multiple times to me tho, always in the same scenario; when a build fails. That is really odd.. I was thinking about the possibility of tmux causing issues, but that seems a bit far fetched.
To be honest always it failed for me due to missing repository content for cmake it never paniced for me, so I'm clueless how exactly this happens.
If you have a public repository with the code / you would invite me read access, I can take a deeper lock into this in September.
Eventually it would be helpful to dump all logging fields in json to the disk, to understand why the logger fall through to the end of all displayed sections. The problem is to recreate the section based log model, logging everything in plain text would be simple and fool proof.
I built the runner from main and ran into a panic when a job fails. Here's some hard- and software information:
The stacktrace is as follows:
Thanks in advance for taking some time to look at this issue :)