03:44:59 #startmeeting Workstation WG (2021-08-31) 03:44:59 Meeting started Wed Sep 1 03:44:59 2021 UTC. 03:44:59 This meeting is logged and archived in a public location. 03:44:59 The chair is cmurf. Information about MeetBot at http://wiki.debian.org/MeetBot. 03:44:59 Useful Commands: #action #agreed #halp #info #idea #link #topic. 03:44:59 The meeting name has been set to 'workstation_wg_(2021-08-31)' 03:45:01 #meetingname workstation 03:45:01 The meeting name has been set to 'workstation' 03:45:03 #chair: Michael Neal 03:45:03 Current chairs: : Michael Neal cmurf 03:45:22 #topic Rollcall 03:45:24 #info present: mclasen, Chris, Owen, Tomas, Jens, Michael Catanzaro, , Neal 03:45:26 #info present guests: Luna Jernberg (bittin), Zac, James, Davide 03:45:28 #topic Approval of August 24 minutes 03:45:30 #link https://meetbot.fedoraproject.org/fedora-meeting-2/2021/fedora-meeting-2.2021-08-24-17.15.html 03:45:32 #agreed Approved 03:45:34 #topic Announcements, follow-ups, status reports 03:45:36 #info Michael: hard code freeze for GNOME 41 starts on saturday (twice as long as usual, as an experiment). 03:45:38 #info GNOME 41 RC Tarballs due on on Saturday 03:45:40 #info Devconf US Online on Thursday-Friday this week 03:45:42 #link https://www.devconf.info/us/ 03:45:44 #topic Make use of malcontent in the default install? 03:45:46 #link https://pagure.io/fedora-workstation/issue/186 03:45:48 #agreed include in F35 Beta 03:45:50 #info Status: Adam waits for us to say: "Yes, put it into the beta". We will do so. 03:45:52 #topic Re-revisit Fedora Workstation minimums https://pagure.io/fedora-workstation/issue/241 03:45:54 - Neal tried with 2G of ram (and other settings simulating a 2009 MacBook pro), and things worked fine. 03:45:56 His theory is that the underlying issue is actually hard-drive performance. Things are fine with an ssd. 03:45:58 - Neal points at lack of readahead (since systemd removed it). 03:46:00 - Chris and Neal suggest that we need to confirm that theory before making changes to the minimum 03:46:02 requirements. 03:46:04 - Owen concurs, and points out that readhead should really only affect boot time, and the 03:46:06 IO scheduler can also influence things. 03:46:08 - Matthias says it would be great to know what percentage of the userbase still uses non-ssd drives. 03:46:10 #info Suggestion from Owen: change the requirements to say "SSD suggested". 03:46:12 #action Chris will file a ticket to request "SSD suggested" being added 03:46:14 #action Neal will follow up with Matthew in the ticket 03:46:16 #topic Improve the user feedback for OOM situations 03:46:18 #link https://pagure.io/fedora-workstation/issue/202 03:46:20 Deferred to a later meeting, since it targets F36. 03:46:22 #topic User directories as Btrfs subvolumes by default 03:46:24 #link https://pagure.io/fedora-workstation/issue/232 03:46:26 - Neal suggests this change, since it will make it easier to use btrfs tools to operate on the whole of the 03:46:28 users data (think backup, compression levels, etc). The target for this is F36 (or F37, depending on getting 03:46:30 bugs fixed). 03:46:32 - Tomas asks if there are any downsides. Neal admits that he can't boot with this configuration :) 03:46:34 But that is just a bug, according to Neal. The other downside is that hardlinks don't work, since inode 03:46:36 numbers don't span across subvolumes. Matthias points out that there was a recent lwn article about 03:46:38 problems between btrfs and nfs wrt to inode numbers. Neal notes this is a problem if you export the parent of subvolumes via NFS, not the subvolumes themselves. 03:46:40 Minor benefits: it becomes harder for containers to cause inode exhaustion. 03:46:42 #link https://lwn.net/Articles/866582/ 03:46:44 #link https://lwn.net/Articles/866709/ 03:46:46 Chris points out that some backup tools don't cross filesystem boundaries by default, and consider subvolumes 03:46:48 as such. An example mentioned by Chris is a tool called Pika Backup. Chris is worried about more knock-on 03:46:50 effects like this, and suggests getting some exposure and file bugs. 03:46:52 The anaconda UI for nested btrfs subvolumes is not great. 03:46:54 #action Neal to file the bugs he found and get them fixed. Then we can revisit this. Chris to help out too 03:46:56 #topic Open Floor 03:46:58 Luna points out that the hard code freeze is Saturday, not Friday as Michael said earlier. and the RC of 41 will be out on Saturday too 03:47:00 #info Jens: devconf.us happens this week 03:47:02 #link https://www.devconf.info/us/ 03:47:44 #endmeeting