I was looking at this idea for GSoC2010: https://issues.apache.org/jira/browse/CASSANDRA-875
I've put a comment on that ticket, but am going to post here as well as it seems to be more active. Sorry for the double post. If this is going to be part of the continuous integration process (assuming Hudson?) what limits would be put on it? How exhaustive should it be? testing only reads/writes? across nodes? I've been playing around with Cassandra a little, but am yet to use all of it's features. How related is this to this ticket: https://issues.apache.org/jira/browse/CASSANDRA-874 If you're planning on making the regression tests part of a CI process, then the virtualised node instances does look impressive. My only concern is that I haven't been able to get it to build. Is this is within the scope of the first ticket I linked to? If so I might e-mail the team from Yahoo about how they've used it successfully. - Kushal
