A Project Involving Socorro
As part of my Bachelor of Engineering at the University of Auckland, I am required to complete a year long research project with a partner. At the end of the project we present our work along with a comprehensive report detailing what we’ve learned. Fortunately my partner is Tony, and Tony had an awesome time interning for Mozilla over the (New Zealand) summer. Using his wit and charm he managed to swing us a Mozilla sponsored project revolving around Socorro, Mozilla’s distributed crash collection system.
What is Socorro?
Socorro is one of those back end things which are critically important, but not particularly glamorous. Before Tony worked on it for his internship, I had never heard of it. Mozilla uses Socorro to collect crash reports from, among other things, Firefox, which with ~450,000,000 users is no small task. To see how Socorro’s architecture is distributed and to get an idea of how it might scale, you can read Tony’s overview of it’s architecture.
What are we doing to Socorro?
It’s currently week 5 of the project, and we’ve spent most of that time deciding and planning our direction for this project. We intend on improving what are some of the wartiest parts of Socorro.
Configurator
The configurator will be a web tool which will help generate and validate Socorro configuration files. Socorro is configured with a collection of INI files which can be hard to keep in sync. Our configuration tool should hopefully make this simpler by generating correct config files from a form based web interface. We hope to be able to export a topology graph of the nodes and storages in the Socorro config, which should make it easier to spot any derps in the set up.
Operations Dashboard
Currently Socorro has a status page which gives information mainly about jobs in the system. We hope to improve on this by creating a dashboard which has information about the system components themselves (the collectors, processors, etc.), including stuff like uptime and load, with the intent of making it easier to debug problems in the system.
Going forward
Keep an eye on this blog and on Tony’s as we go forward with this project. The final deliverable is due in September, so until then expect fairly frequent updates between us about our progress.