As we do every two weeks, we released new improvements to the Ohloh software today. We dedicated this entire release to performance and reliability improvements. So while there are no new visible features for us to promote, there are important improvements behind the scenes that should lead to faster updates and fewer problems for end users.
Ohloh operates a large cluster of servers that together process over 50,000 source code downloads and analyses every day. With so much going on, if even just a small percentage of our jobs fail, we end up with a large number of problems that need to be investigated.
The new Ohloh release contains better back end tools to help us manage those failures. By categorizing similar failures, we can identify, fix, and re-schedule jobs more efficiently. This should lead to fewer noticeable problems for our users.
We’ve also instrumented the Ohloh server code with profiling hooks to measure our performance at every step during our analyses. This will help us identify the hot spots in Ohloh’s server code, telling us where to focus our attention to speed up Ohloh’s processing overall.
The improvements won’t be obvious immediately, but we’re dedicated to improving the speed and accuracy of Ohloh over the long term. Thanks for supporting Ohloh, and as always, we look forward to hearing your feedback.