@chadoh I edited the sandbox install step to delete the file and try again, but with debug info. This seems to cause the test to pass the second go around, which points to a concurrency issue. Since all the test files are all checking to see if the binary exists and then trying to install it, there is definitely a race condition. So my guess is that one of the tests incorrectly sees the binary as ready and tries for it. We could add one more check right before executing the binary for the first time. However, the good news is that the post install step seems to work fine.
@chadoh I edited the sandbox install step to delete the file and try again, but with debug info. This seems to cause the test to pass the second go around, which points to a concurrency issue. Since all the test files are all checking to see if the binary exists and then trying to install it, there is definitely a race condition. So my guess is that one of the tests incorrectly sees the binary as ready and tries for it. We could add one more check right before executing the binary for the first time. However, the good news is that the post install step seems to work fine.