[OpenAFS-devel] afs test suite

[Sam Hartman]

>>>>> "Derrick" == Derrick J Brashear <shadow@dementia.org> writes:

    Derrick> Is anyone willing to and interested in working on a test
    Derrick> suite? If someone is interested a draft of requirements
    Derrick> for much a suite can be made up.  Arla has a test suite
    Derrick> which can be adapted, but it's not broad enough to cover
    Derrick> everything which should be tested (though it does a good
    Derrick> job of testing file system semantics).

As we start picking up customers, Mekinok is definitely going to need
to spend effort on this.

    Derrick> My proposal would be something like: -assume a complete
    Derrick> cell available for testing purposes -o

We'd probably want to create the cell as part of the test.  This is
because we want a completely reproducable environment and because cell
creation is something we care about a lot.  Also, it will probably
need to involve multiple clients to test and make sure that cross-host
locking and cache consistency guarantees are maintained.

I have some general ideas on framework I'll try and write up over the
next few days.  I agree that starting small with tests that assume an
existing cell would probably be wise, but we want to design the
framework to meet more complex needs long-term.