WARNING: This is the _old_ Lustre wiki, and it is in the process of being retired. The information found here is all likely to be out of date. Please search the new wiki for more up to date information.
Testing Lustre Code
Create test at beginning of development process – test often as you develop code - Overview of testing framework (acceptance-small.sh, etc.)
Start your work with the development of a test to reproduce the problem or test the new feature. This allows you to verify that the problem you are working to fix is the same one hit by the customer.
This will also allow you to verify that the change you make is actually fixing the problem. You can add failure hooks with the OBD_FAIL_CHECK() or OBD_FAIL_TIMEOUT() hooks in Lustre if needed.
For details, consult Testing Framework.
There is a script in lustre/tests called "acceptance-small.sh", which can automatically run a bunch of tests.
Make sure acceptance-small.sh passes on your workstation. It can be run directly from the lustre/tests directory and it is a good idea to run it often.
This allows you to know as soon as possible if your code is adding a regression.
Writing regression tests early to reproduce a bug will allow you to confirm you are hitting the bug.
Having a specific regression test before fixing the bug will allow you to verify that the bug is actually fixed.
This saves you the effort of testing a fix manually AND writing a regression test. Your code must pass acceptance-small you commit it to CVS.
Please set the "acc-sm passed" flag on the attachment for each individual branch that was tested in Bugzilla to indicate this.
