mirror of
https://github.com/etcd-io/etcd.git
synced 2024-09-27 06:25:44 +00:00

The test is vary flaky on Travis. Seems that since (https://github.com/etcd-io/etcd/issues/7724) the client is expected to simply ignore whether server is in AuthDisabled mode even if the user supplies credentials. The tests used to: * use very large cluster (10 nodes) * set very low timeout (1 sec) Such setup led to frequent deadlineExceed errors or following failures: === RUN TestGetTokenWithoutAuth {"level":"warn","ts":"2020-08-04T16:50:48.686+0200","caller":"clientv3/retry_interceptor.go:62","msg":"retrying of unary invoker failed","target":"endpoint://client-35573307-1ee5-441b-acc7-d073f0bd7de5/localhost:69820396562031027440","attempt":0,"error":"rpc error: code = Unavailable desc = etcdserver: leader changed"} user_test.go:151: other errors:etcdserver: leader changed --- FAIL: TestGetTokenWithoutAuth (10.91s)