Fedora Test Day – recap

So, a few days later … I’ve had some time to think.

My first Fedora Test Day was a busy event. I’m very pleased with the participation level. Several new users joined the event, which was a great experience for me. I’m thankful for their time! Especially when it comes to helping review and execute test documentation. A fresh set of eyes can often point out things you’ve glossed over or grown accustomed to workingaround. Their contribution also helped highlight the ubiquity of virtualization. While I’ve seen it before in earlier discussion in #fedora-qa, and worked with it for many years on ppc, it didn’t resonate until experienced first hand. Whether it’s qemu, kvm, xen, or VirtualBox or $other … virtualization has become a household name.

Having developers Ray Strode (plymouth) and Dave Lehman (anaconda) available for Q+A for much of the event was instrumental. As the key folks responsible for two very visible features in Fedora 10, what more can you ask for? With no boundaries in the way (bugzilla, mailing lists), we were able to engage in great collaborative discussion around known issues, expectations and use cases. Out of that came a list of bugs and enhancement requests (some of which were patched the same day). The list included:

Bug 459116 anaconda needs to restrict certain encrypted device combinations
Bug 459094 Password prompt for encrypted devices doesn’t show which device is about to decrypt
Bug 459115 Plymouth fails to decrypt secondary partitions from the first time
Bug 459121 AttributeError: RaidRequestSpec instance has no attribute ‘setPassphrase’
Bug 459123 Encrypted Software RAID device members don’t display a lock icon
Bug 459191 RFE – plymouth boot support for global passphrase
Bug 459089 Entering wrong passphrase when attempting to disk probe continues to prompt for passphrase without imforming the user the password is wrong
Bug 458278 Plymouth should not fail on headless systems (serial console)
Bug 459111 When prompted for passphrase to unlock root device, hitting <esc> prevents passphrase entry

While overall a very positive experience, these things wouldn’t be worthwhile if there weren’t improvements to be made. Some things that proved a challenge:

  • Plymouth, while very promising once you have things lined up, was a bit rough around the edges for alpha. Newer versions of plymouth built after F10-Alpha show much more promise, and the soon-to-land kernel modeline support will remove the need for booting with "vga=0x318". Until Beta trees begin to land, you can test out the latest plymouth by upgrading from rawhide:
    $ yum update plymouth mkinitrd
    $ mkinitrd -f /boot/initrd-$(uname -r) $(uname -r)
    $ grubby --update-kernel=/boot/vmlinuz-$(uname -r) --args "rhgb vga=0x318"
  • Testing encrypted installations requires one to think.  While perhaps obvious, it’s too easy to have 3 to 4 different entries in the test matrix going and then forgetting which is which when that all go south.  There are many combinations to test and the support for stacking encryption doesn’t ease the burden.  How are you using encrypted block devices on your systems?
  • If you don’t like typing a passphrase, you’re in trouble.  Typing in the passphrase to unlock previously encrypted devices, to configure newly encrypted block devices, and to unlock them during bootup  … is not only frustrating, but your hands will begin to hurt after typing the same stuff that many times.  Ray thought of that.  Discussion turned towards enhancing plymouth to remember your passphrase during boot (see Bug 459191).  Dave also is debating extending the global passphrase support when probing existing devices to be used during device creation.  Both will be barely visible changes, but much appreciated usability enhancements.
  • Testing installation issues can be easier.  While the Live image (CD/USB) are extremely useful for rapidly exercising user-interface functionality, there still are a lot of moving pieces.  Whether it’s rawhide, a snapshot, a milestone, a yum repo, a compose tree, or a set of iso’s, build your own, or pull it from a mirror.  All the pieces are out there.  But how do they fit together?

In short … thanks everyone for participating!  I look forward to another exciting Fedora Test Day [1] walking through a few more Fedora 10 Accepted Features.

[1] Alternative names submissions being accepted.  The current candidate proposed by Robert Williams: TestFest!

Advertisements

2 responses to “Fedora Test Day – recap

  1. Cool 🙂

    Hey,

    Just wanted to say this is a cool project, and I hope you repeat it soon. I was really sorry not to be able to attend the day myself, but now that I’m back home from travels for the time being I’ll definitely make the next one…

    Again, great work!

    JonRob

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s