Brief Litmus Downtime Tonight
Posted 1 year, 8 months ago at 10:35. 0 comments
Mozilla IT is performing some maintenance tonight, and Litmus is affected. Downtime should be minimal.
You are currently browsing the archives for the Litmus category.
Mozilla IT is performing some maintenance tonight, and Litmus is affected. Downtime should be minimal.
The scheduled downtime never actually happened last Thursday, so we’re going to try again tomorrow night.
We’re taking Litmus down briefly this coming Thursday afternoon (Feb. 8th, 2007) to install the VMware tools on the VM. Among other things, this will allow proper time syncing from the host to the VM.
We don’t anticipate an extended outage, but we’re budgeting for an hour from 5pm - 6pm PST just to be safe.
I took advantage of the US holiday yesterday to update the Litmus wiki with the current state of affairs.
I’ve added a top-level Work in Progress section so people can tell at a glance what we (Zach and I) are working on now. More detail is still available in the To-Do List. I’ve also made the Wishlist more obvious. I’ll try to add the existing enhancement request bugs to the wishlist shortly.
There are lots of other little changes: rudimentary breadcrumbs, links to bug lists, and my personal favorite, a stub for a section on Selenium tests for Litmus. I’ve got a small suite of them running now, and can’t wait to add more.
We’re changing things up a little bit this week. Our bi-weekly testday is returning after a holiday hiatus, and this time we’re going to be focused on improving test case coverage either by writing new test cases, or by shoring up existing ones.
More information about this testday can be found in the Mozilla wiki.
Axel has even offered to show up to help any l10n contributors improve our somewhat-woeful l10n test case situation.
Hope to see everyone in IRC!
After having a conversation with Tomcat this morning, I just wanted to clear up a potential point of confusion for our community testers in relation to the testing coverage statistics we provide in Litmus. The chances of seeing high coverage values for subgroups in Litmus increases on testdays when our testing focus is particularly targeted, but there are still several ways to provide helpful testing feedback under those circumstances.
Here’s a simple plan of attack:
I just landed the changes for bug 355278.
These changes address a long-standing underlying architectural problem in Litmus whereby subgroups and testcases were only associated with a given code branch (e.g. Trunk) if they were already contained in a testgroup that belonged to that branch. This presented a chicken-and-egg problem to admins who were trying to add new subgroups or testcases. A workaround did exist, but it wasn’t readily apparent, at least not based on the number of questions I fielded about it.
The solution was to associate subgroups and testcases with branches directly. This does mean that, going forward, testcases will need to be cloned rather than re-used for each branch. In most cases a given testable feature will show some delta between major versions (i.e. branches), and this was how we at Mozilla were using the system anyway.
If you’re running your own Litmus install out there, I’ve provided a migration script in migration/update_branch_relationships.pl, that will do the heavy lifting in terms of mapping existing testgroup->branch relationships over to subgroups and testcases.
Lots of discussion and decision-making last week, only some of which was fueled by well-deserved booze. Here’s a wrap-up of the salient points as they pertain to me.
Build/Release:
QA:
Litmus:
Our most recent community testday was a resounding success.
We had around 35 people actively testing, many of whom were joining us for the first time. This was largely due to our collaboration with Seneca college in Toronto. Thanks to all the Seneca students who showed up to help with testing, and thanks to robcee for leading that particular charge.
The testday generated over 1000 results. For those keeping score at home, that’s triple our previous record for testday result submission. We got lots of good coverage on Linux (which is rare for us on testday), and also received lots of valuable feedback on testcases and on Litmus itself. So much feedback, in fact, that I needed to whip up a quick interface to view results with comments in the middle of testday. We also debuted the new report interface for testday statistics after the testday.
Thanks again to everyone who helped make this past testday a success. We hope you’ll decide to come back and join us for the next one.
Aravind got a new VM setup for us, and Zach did the heavy lifting while I was away on vacation this week to get Litmus working right using a much more up-to-date version of MySQL and Apache. Let testing commence anew!
Litmus performance has been a bit of a sore point as of late, and we’ve been meaning to move into a VM for a while, so I would be interested to hear people’s observations on the responsiveness of the new setup.