qserv-admin.py is somewhat geared towards interactive usage. Upon startup, it can be fed a sequence of commands through stdin, and it attempts to execute them, printing to stdout if there are any errors. This is perfectly fine for humans running qserv-admin directly.

But qserv-admin.py is also called from the integration tester. The integration tester doesn't have any error parsing skills, so it doesn't check stdout. I think it may check an error code . Thus the integration tester has no idea whether qserv-admin.py commands succeed or not.

I propose adding a switch to qserv-admin.py that instructs it to abort with an error code on any error. I'd also like to know whether the integration tester can understand error codes (I think so, but am not sure).