Make test of Cirrus in any cluster
jcarreira opened this issue · comments
Right now some of our tests make some assumptions specific to our development environment (e.g., IPs of servers).
We should allow make test to run anywhere.
The current setup is that the "make test" command automatically runs any scripts listed under the Tests variable in the highest level makefile.ac . As such, this means that there is no real way to pass arguments into the test scripts, which then call the tests themselves. One solution could be to create a configuration file that resides in the tests directory. The scripts that launch the executables could then read from this config file and pass the arguments (ip) to the tests themselves.
Another factor is that right now the tests work by starting the server locally, running the test, killing the server, and then verifying exit status. I'm assuming there is no way for a python script to launch a server on another machine, so perhaps to accommodate remote testing the only change we have to have here is have the configuration file be of the format "ip XXXXXX is_local BOOL". With this format, the launcher script would be able to know whether or not to launch the server locally, and instead assume that it is running remotely already.
Could this be merged with #51 ?
I'm going to close this in favor of #51