| xnox | cjwatson: stgraber: elmo says /boot should be more than 256MB in 2014-2019 =) i think we should give /boot same or more than EFI. | 15:28 |
|---|---|---|
| xnox | efi is at: 512 512 1024 free | 15:28 |
| xnox | and | 15:29 |
| xnox | boot is at: | 15:29 |
| xnox | 128 512 256 ext2 | 15:29 |
| cjwatson | ideally, in 2014-2019, /boot shouldn't need to exist ... | 15:31 |
| cjwatson | but sure, feel free to bump the max size | 15:31 |
| cjwatson | EFI is only that large because of stupid fat32 constraints, though - don't take it as any kind of specific guideline | 15:31 |
| infinity | My /boot only has 81MB in it... | 15:42 |
| infinity | I suppose actually autoremoving helps. | 15:42 |
| xnox | infinity: how often / when is autoremoval is triggered on the default installs? | 15:50 |
| infinity | xnox: Triggered? Never. | 15:57 |
| infinity | xnox: That's probably something both unattended-upgrades and update-manager should learn to do, BUT, we need to argue about how to do it safely. | 15:58 |
| xnox | infinity: so how does marking kernels for autoremoval help, if autoremoval never happens (e.g. neither by default apt-get, periodic apt-get, update-manager / unattended-upgrades) ?! | 15:58 |
| infinity | xnox: It helps for people who use apt-get autoremove. | 15:59 |
| infinity | Pretty sure no one said it was perfect, nor that questioning it required flailing and interrobangs. | 15:59 |
| infinity | The deep concern with automatic autoremoval is the potention for people to shoot themselves in the foot. | 16:00 |
| infinity | But maybe that's just a question of "people should learn how metapackages work and not whine when they break stuff". | 16:00 |
| infinity | Since the most obvious breakage would be "apt-get install corp-metapackage; remove corp-metapackage; (automated tool removes all of corp's packages; corp files angry bug and escalates support ticket and threatens to eat our children)" | 16:01 |
| xnox | infinity: that's no different then a corp's user doing $ apt-get autoremove or doing --auto-remove. etc. if corp wants it installed, it should keep it installed, i heard puppet is good at installing things when they are removed. | 16:06 |
| xnox | infinity: imho /etc/apt/apt.conf.d/50unattended-upgrades should change to Unattended-Upgrade::Remove-Unused-Dependencies "true"; | 16:07 |
| infinity | Having something happen automatically is definitely more surprising than a user explicitly calling autoremove. | 16:08 |
| infinity | I'm not saying it's wrong, I'm jut saying we need to weigh the options and decide if it's an issue. | 16:08 |
| antarus | infinity: yeah we wrote our own kernel remover, the debian folks on the team seemed to massively distrust apt-get autoremove for some reason ;p | 20:20 |
| infinity | antarus: I had a healthy distrust for it a long while back, it all seems to work as advertised now. | 20:22 |
| infinity | It's certainly not bitten me in any unexpected ways in a very long time. | 20:23 |
| antarus | yeah, I think her fear is somewhat justified | 20:23 |
| antarus | mostly because I bet our end users do some crazy stuff | 20:23 |
| antarus | ;p | 20:23 |
| infinity | Users will be users. | 20:24 |
| antarus | what I really want is something more like apt-get autoremove <depstring> | 20:24 |
| antarus | instead of acting on all packages | 20:24 |
| antarus | but I don't think that exists | 20:24 |
| === YamakasY_ is now known as YamakasY | ||
| === TheMuso` is now known as TheMuso | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!