Skip to main content

... and you have to CI your Jira too!

Posted by fabriziogiudici on August 19, 2009 at 1:18 PM PDT

I've just received a patch for jrawio from a user. Great stuff. But the user complained about the fact he could log in on my Jira (jrawio uses my own installation at Tidalwave), but he couldn't post comments or attach the patch. He's right in complaining, of course. Usually I pay a lot of care with the communication infrastructure for a project, and in the past there were no specific problems with my Jira installation. Now, as I'm changing the world (forge, web site, whatever) I've probably made some misconfiguration. Not a big deal, I'll fix it - but this sounds as the old rule "if you didn't test it, it's broken". Since in the past Jira worked, I assumed it was still working. My bigger mistake is that I wasn't probing it continuously (of course, I'm using Jira all the days, but clearly the problem is not visible with my account).

So, the question is: should I set up some sort of automatic, continuous testing of the infrastructure? Do you do something similar? How?


Well, consider two things: 1) I try to keep my open source projects in the best possible way (compatible with the very limited budget), also to experiment things to later use for industrial projects or for teaching stuff to people, considering that I can't disclose stuff from other customers' projects 2) if you have an open project such as jrawio, that is fundamental for certain users, but in a small niche, and it's hard to aggregate people working on it, it's no good to receive complains from people that would like to participate, but were blocked by some sort of problem.

I've set up system monitoring before but it was for systems that were critical for production (as in "we're going to lose money during downtime.") It seems like overkill for an open source project, though maybe warranted if the infrastructure is inherently unstable.