What I'd like to work on
Fix for issue #4321 - OTEL_EXPORTER_OTLP_ENDPOINT in the environment breaks trigger deploy (depot build).
I've already written the fix and opened PR #4370. The change is in packages/cli-v3/src/deploy/buildImage.ts - adding OTEL_SDK_DISABLED: "1" and OTEL_EXPORTER_OTLP_ENDPOINT: "" to the env block passed to the depot child process, so ambient OTEL config from the parent shell can't reach it. 7 lines total (2 code + 5 comment explaining why).
Background
CS undergrad, been doing open source contributions seriously for the past few months. 75 merged PRs across 104 GitHub orgs. Recent merges include redis/redis-py (5 PRs, test coverage for cluster hash-tag logic and utility helpers), numpy/numpy (thread-safety fix for np.dot under concurrent load), and a few smaller projects.
I found this issue by reading through the open bugs list and traced the root cause in the source directly - the depot() call uses execa's default extendEnv: true, so whatever the parent shell has in OTEL_EXPORTER_OTLP_ENDPOINT leaks into the depot child's telemetry init and kills the build before it starts.
Happy to adjust anything in the PR if the approach isn't what you'd want.
What I'd like to work on
Fix for issue #4321 -
OTEL_EXPORTER_OTLP_ENDPOINTin the environment breakstrigger deploy(depot build).I've already written the fix and opened PR #4370. The change is in
packages/cli-v3/src/deploy/buildImage.ts- addingOTEL_SDK_DISABLED: "1"andOTEL_EXPORTER_OTLP_ENDPOINT: ""to the env block passed to the depot child process, so ambient OTEL config from the parent shell can't reach it. 7 lines total (2 code + 5 comment explaining why).Background
CS undergrad, been doing open source contributions seriously for the past few months. 75 merged PRs across 104 GitHub orgs. Recent merges include redis/redis-py (5 PRs, test coverage for cluster hash-tag logic and utility helpers), numpy/numpy (thread-safety fix for np.dot under concurrent load), and a few smaller projects.
I found this issue by reading through the open bugs list and traced the root cause in the source directly - the depot() call uses execa's default extendEnv: true, so whatever the parent shell has in OTEL_EXPORTER_OTLP_ENDPOINT leaks into the depot child's telemetry init and kills the build before it starts.
Happy to adjust anything in the PR if the approach isn't what you'd want.