search

kids-first / kf-model-fhir

🔥 FHIR Data Model for Kids First
https://kids-first.github.io/kf-model-fhir/ig/
Apache License 2.0
1 stars 1 forks source link

✨ Add genomic file target api builder #202

Closed liberaliscomputing closed 4 years ago

liberaliscomputing commented 4 years ago

This PR:

It closes #189.

liberaliscomputing commented 4 years ago

The below CI tests look weird. In fact, the results from Jenkins and CircleCI conflict: Circle's result is the same as local integration testing which has got passed.

znatty22 commented 4 years ago

Ok @liberaliscomputing looking into it now

znatty22 commented 4 years ago

Ok @liberaliscomputing looks like tests are passing.

I think what might have happened is the integration tests started during build 1, deployed a fresh temp integration test server, started loading, then exited due to an error - before getting to the part in the integration_test.sh script where its supposed to delete the temp integration test server. Then in subsequent builds, the integration_test.sh script sees a server running and doesn't run a new one, but there was probably old data in there causing the count tests to fail. I manually deleted the integration test server on Jenkins so that the integration_script.sh.

I might update this script so that it's more resilient to this sort of thing.

liberaliscomputing commented 4 years ago

Ok @liberaliscomputing looks like tests are passing.

I think what might have happened is the integration tests started during build 1, deployed a fresh temp integration test server, started loading, then exited due to an error - before getting to the part in the integration_test.sh script where its supposed to delete the temp integration test server. Then in subsequent builds, the integration_test.sh script sees a server running and doesn't run a new one, but there was probably old data in there causing the count tests to fail. I manually deleted the integration test server on Jenkins so that the integration_script.sh.

I might update this script so that it's more resilient to this sort of thing.

@znatty22, ah that makes sense totally! So, it failed because 4 != 2 from Organizations and the additional two were lingering from the previous build! Yeah, I remember I force-pushed a minor tweak during the first test was running. Thanks for digging into it!