David Steinberg
David Steinberg
@jeromekelleher I believe you are correct, someone was using the swagger docs during the night! Future load tests will be against a private server and performed in a more coordinated...
Unfortunately, it's something I wasn't quite able to google for. If someone knows of a realistic multi-contig VCF a link would be helpful!
Related https://github.com/ga4gh/schemas/issues/828
Verified by turning back on the test. `htslib` @jmarshall crashed with: The VCF in question: https://github.com/ga4gh/server/tree/master/tests/faultydata/variants/duplicated_sampleid ``` (env)➜ server-1 git:(1253_codecov) ✗ nosetests tests.unit.test_faultydata [E::bcf_hdr_add_sample] Duplicated sample name 'S1' [1] 39488...
Issue opened, we'll leave this here to help track progress on the htslib side.
Big step towards this with removing the converters. I think `convert_error_code` should not live with the server.
I believe this specific error logging case is covered by https://github.com/ga4gh/server/pull/1154.
In action here: https://github.com/ga4gh/server/blob/1b1df9809ebe9a9bc6281a00de6ded9f90d43268/ga4gh/frontend.py#L829
To better demonstrate what this gives us, I ran the locust load testing suite with this locustfile against this PR in two modes. This test is a little better isolated...
If we merge this we should follow it with a PR that improves test coverage under gunicorn mode in test_gestalt. Tests that involve ensuring an announce request is sent and...