Apple tart and snow in Amsterdam

Southern Europe is suffering a very cold spell indeed. If I walk outside now in Amsterdam then the temperature is -4 with a wind chill factor that makes for effective temperatures below -10. The weather is great for the children as car driver’s ride slowly by only to get snow balled (Alan say's as he hides a snowball behind his back).

The cold weather motivates me to sit for longer and warmer behind the computer. However, luckily I am taking time to eat home made apple tart and drink warm drinks of various levels of non alcoholic content, hic.

Over Christmas not only did I focus on quality time with the family, I also made some time to think out some prototypes for QA automation and started to use Findbugs a tool for yeah finding bugs in compiled Java code to search for near surface defects in Sakai 2-7-0-M2. I was especially interested in resource leaks and things that can generate error messages for end users. The static code sweep is documented at http://jira.sakaiproject.org/browse/SAK-17647. The success of the sweep depends on the hard working developers cleaning the code in a timely manner. The worst thing that can happen is that defects are brought to the surface to only be ignored. Luckily, a lot of work has already taken. For example, David Horwitz has actively cleaned a number of potential resource leaks in the kernel and search. Each defect in itself may feel trivial, but taken as a whole help with the stability of Sakai under high load.

Just before Christmas Nuno Fernandes and Chris Kretler wrote and performed grinder stress testing for the sitestats tools and managed to heat up qa1-nl considerably. That was fun and hinted at places for the QA to focus effort (not related to sitestats). Nuno Fernandes has agreed to perform the tests again after the bulk of the static code review defects mentioned have been removed.

This month is a busy month for all those involved in QA. We want to focus on the 2.7 tag series, but still need to double check 2.6.2 and 2.5.6 before the official tags are released. Minimally, the conversion scripts need to be verified and this requires QA admin’s to promptly bring up various versions of Sakai converting Oracle and MySQL databases with data and then QA need to perform a number of holistic tests.

Seth Theriault is using a QA inspired stress testing script to stress test http://qa2-us.sakaiproject.org:9090 for longer periods of time at about 500,000 hits per day. This Jmeter script has run over Christmas and has generated no significant errors. This gives a good baseline test and I intend to use the same scripts to perform a sanity check on 2.6.2 and 2.5.6 just before they go out the door.

On the automation front, http://builds.sakaiproject.org:8080 is a hudson continuous build server. I expect this will start to play a significant role in QA’ing in the coming year; however we are presently waiting for a memory upgrade from 2GB-> 4GB before being able to run the more demanding features.

Corey McGarrahan who works for rSmart has been kindly leading a significant effort to create automated functional tests via selenium-rc. As the work matures, I expect to make more announcements.

Sakai 3 is on the horizon for the community, therefore I am also starting to look at how central QA can involve itself productively.

Finally, a little belated, but happy new year to one and all, I am now going outside to be hit by snowballs. I will go down fighting and then back to the work with a smile on my face and a limp in my leg.