ci: Resolving nginx & prometheus build issue + web server consistency - #633
ci: Resolving nginx & prometheus build issue + web server consistency#633thompson-tomo wants to merge 0 commit into
Conversation
a4b63bb to
4d12248
Compare
e9eaaab to
2b750a0
Compare
|
Thanks for improving the CI here - most of the previously failing jobs are now green. One Prometheus issue remains: the Linux CMake job still fails while configuring the upstream |
|
@lalitb i am out of ideas as to what to do to get it green given my knowledge of these types of builds. |
@thompson-tomo - The Prometheus workflow performs two separate CMake configurations:
The failure happens during the first configuration, when This PR adds Could you make ZLIB available before the first CMake command that enables Prometheus? |
|
@thompson-tomo |
|
@proost i would suggest you cherrypick the changes to https://github.com/thompson-tomo/opentelemetry-cpp-contrib/blob/ci_fix/.github/workflows/webserver.yml onto your branch which will enable the ci to run in your branch. I will then remove them from here to avoid conflicts. |
|
@lalitb / @dbarker any further guidance on how to resolve https://github.com/open-telemetry/opentelemetry-cpp-contrib/actions/runs/32260993486/job/96093945611#step:5:1 |
|
Please see #667 for continuation of this pr |
NGINX instrumentation build failures have been addressed by explicitly using buildx & switching tests to use php docker image which exists. Have also updated the tests to use nil for remote-parent-id rather than empty string.
Prometheus bazel build resolved by specifying an older bazel version to use.
Prometheus cmake build resolved by changed zlib adding process.
Httpd/webserver resolved by updating runner image.
This restricts all push triggers to main to improve ci consistency and remove unnecessary checks on pr’s. To build for a commit use the manual trigger or use a draft pr.