Comment from #153:
The number of people testing Fedora with btrfs on their main machines is currently extremely small. We think we'll have more people testing this over the next few months with the decision to switch to btrfs. Is this going to just be a some FESCO and WG members, or will we succeed in getting hundreds of people testing this? How will we know?
One of the goals here is to ascertain the number of people testing btrfs, so we can correctly judge how effective our testing efforts have been. We've discussed having a form that people can use to sign up to the testing programme, and possibly following up with them to find out if they've had any issues.
I'm assuming that we'd have to figure out how to navigate GDPR if we want to do this.
@sumantrom would like to write an Fedora Magazine article to get more testers for the next test day (more like test week). This would be in a style similar to those used to announce+recruit testers for kernel test days.
The first test day results - note that these test cases are part of the on-going validation requirements for Fedora releases. These tests are going to happen again and again, and will get filled in by testers for regularly nominated test images, and by many extra testers for beta and final release candidates. The first btrfs test day was essentially a run through of what we'd see for a typical (desktop) release candidate. And to find early breakage - and it did that, despite the overwhelming green checkmarks. Most breakage was in the compose process. https://testdays.fedorainfracloud.org/events/88
But now what do we want to get out of a 2nd test week with this expanded pool of testers? Is the theme recruiting early adopters? User to user support folks? Are there other test cases we can employ from outside of Fedora QA?
cc: @adamwill @kparal
We could ask people to perform some advanced filesystem operations, like resizing, installing btrfs to multiple disks, rescuing a system, running on a hdd instead of a ssd, etc. But in some of these cases, we already identified problems, so it would be nice to have them resolved first, so that people don't just all hit the same obstacle. And those will not get fixed in such a short time, I guess. Also note that the first test day had some test cases related to this, but the response rate was very low in there.
Getting larger numbers of early adopters would be very helpful. Yesterday I read a blog post from Colin Walters about how his Firefox instance is suspiciously slow on btrfs. This is exactly the feedback we want to collect, but it will require weeks or possibly months of usage. So it might be useful now, and it might be even more useful if we delay the change to F34 (assuming we can track all the feedback and it doesn't get just reported into bugzilla and no one looks at it).
But in some of these cases, we already identified problems, so it would be nice to have them resolved first, so that people don't just all hit the same obstacle. And those will not get fixed in such a short time, I guess. Also note that the first test day had some test cases related to this, but the response rate was very low in there.
Contrary to your expectation (thankfully!), @josef and @dcavalca are working on those things. I do expect improvements to land throughout F33 development cycle. That's why we wanted to start this as early as possible.
So it might be useful now, and it might be even more useful if we delay the change to F34 (assuming we can track all the feedback and it doesn't get just reported into bugzilla and no one looks at it).
The only reason we're getting this info now is because we're switching the default. People just don't care unless we do it. And I'm optimistic that we can do things to improve performance. @chrismurphy already identified something suspicious caused by our usage of BFQ I/O scheduler and that's being looked at as well.
This ticket was intended to track a specific testing initiative:
I've started to look into the privacy and legal requirements for this and will report back once I have answers.
This seems obsolete. Since we have released F33 with btrfs by default earlier today, we can assume that almost all newly-installed systems will use btrfs. Can we close this?
Alternative proposal: close this issue and open it under https://pagure.io/fedora-btrfs/project/issues, so it's no longer tracked on the WG issue tracker.
I think we can consider this obsolete now.
Metadata Update from @ngompa: - Issue close_status updated to: Won't fix - Issue status updated to: Closed (was: Open)
Login to comment on this ticket.