In normal conditions, if the http API cannot bootstrap (due to the gRPC server being temporarily unavailable) it will just retry until it connects.
However, this is done asynchronously, meaning that it could be the case that the tower is reported as ready when it is actually not. This can be hit during E2E testing if the tower is restarted too quickly, resulting in a test failure:
In normal conditions, if the http API cannot bootstrap (due to the gRPC server being temporarily unavailable) it will just retry until it connects.
However, this is done asynchronously, meaning that it could be the case that the tower is reported as ready when it is actually not. This can be hit during E2E testing if the tower is restarted too quickly, resulting in a test failure:
https://github.com/sr-gi/rust-teos/runs/8074272734?check_suite_focus=true#step:9:5543 https://github.com/sr-gi/rust-teos/runs/8074272734?check_suite_focus=true#step:9:6970