Socorro Project Update

Mike Morgan

1

There has been a lot going on for the Mozilla crash reporting system. Here are some quick links for project activity:

Upcoming changes:

  • Socorro site redesign – Neil and Austin are working with Sam and Chris to work through crash analysis to understand how the system is used to track down the cause of certain crashes. Structure, navigation and layout will be redone to best accommodate common uses.
  • Socorro database repartitioning and migration – Lars is working hard on refactoring our partitioning to better support the kinds of queries we make. The end result is better long-term maintainability and improved performance.
  • New reports – MTBF reports, URL reports – Austin is working on two new report types to provide better information for crash analysis.

Current issues:

  • Query timeouts – this usually happens when a user does a complex query (either no product selected or a search over a period longer than 4 weeks).
  • There is no dialog saying the app is working and connections time-out leading a user to simply hit refresh to try to get where they want to go. However, many of the queries taxing the system were non-specific queries over 3-month periods. In some cases these jobs would timeout in the reporter but continue for 30 minutes or more in the database. To prevent this from happening we pushed a revision to the search form to limit query breadth.
  • Search engines were crawling Socorro so we added a robots.txt file to prevent this from happening

Learn more:

One response

  1. Ted Mielczarek wrote on ::

    With the new PHP code, it’s remarkably easy to setup a local Socorro for hacking. Setting up Postgresql is a little bit of a pain (but not terrible), but once you have Postgres, Apache, and PHP installed, you just check out the code and edit a few config files (all in the README), and it works!