=== RikMills__ is now known as RikMills | ||
ahasenack | good morning | 11:51 |
---|---|---|
sforshee | xnox, apw: I've got new wireless-regdb packages if one of you could sponser them when you get a chance: https://launchpad.net/~sforshee/+archive/ubuntu/wireless-regdb | 13:23 |
utkarsh2102 | rbasak: hey, I am going to take a look today and tomorrow and sort this out! | 13:40 |
utkarsh2102 | sorry for the delay, this is next in my list of things to do | 13:40 |
rbasak | utkarsh2102: I've already been looking | 13:54 |
rbasak | utkarsh2102: so leave it for now please. I think it's a regression in MySQL 8.0.28 and I'm mostly through writing a reproducer. | 13:55 |
schopin | Anyone know of any documentation on how to package a new upstream version using git-ubuntu (if that's supported?) | 13:57 |
rbasak | schopin: I don't think the git-ubuntu CLI has a specific tool to automated it for you, if that's what you're asking. | 14:02 |
rbasak | The importer should accept the result in the normal way though. | 14:02 |
=== genii-core is now known as genii | ||
schopin | ahasenack: thanks for keeping me in the loop for freeradius, much appreciated. Is the EAP-MD5 issue tracked anywhere? | 16:20 |
ahasenack | schopin: plain eap-md5 works | 16:21 |
ahasenack | it's when used inside ttls that the client I selected for testing refused to accept it | 16:21 |
ahasenack | I was just in copy & paste mode by then | 16:21 |
ahasenack | schopin: the freeradius PR is up, https://code.launchpad.net/~ahasenack/ubuntu/+source/freeradius/+git/freeradius/+merge/415870 | 16:22 |
ahasenack | schopin: eap-md5 through ttls was already like this with my 3.0.25+patches build, I just didn't check another ubuntu release to be sure it's a change in behavior | 16:32 |
ahasenack | there is time for that after FF | 16:32 |
schopin | ahasenack: BTW, are you expecting me to review the MP? Given the size of it, I'm not sure I'd be able to get to it before FF | 16:40 |
ahasenack | if you want, but no, not expecting | 16:51 |
ahasenack | you could take a glance, see if you spot anything obviously wrong, but really it's the upstream snapshot, no extra patches | 16:51 |
ahasenack | in fact, I dropped the patches I had added to 3.0.25 to get it to work/build with openssl3 | 16:52 |
bluca | anybody from the kernel team around? it would be really great to have CONFIG_WATCHDOG_HRTIMER_PRETIMEOUT enabled for Jammy | 17:04 |
bluca | https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1961771 | 17:04 |
ubottu | Launchpad bug 1961771 in linux (Ubuntu) "Enable CONFIG_WATCHDOG_HRTIMER_PRETIMEOUT in Jammy" [Undecided, Confirmed] | 17:04 |
bluca | I'm looking to add a watchdog test for systemd upstream autopkgtest, and we'd use that feature when we rebase to jammy | 17:05 |
ahasenack | what happens in terms of component mismatches when a package in main has a Depends with "a | b"? Are both "a" and "b" required to be in main? | 18:08 |
ahasenack | in this example, "b" is in main | 18:08 |
ahasenack | Depends: | 18:08 |
ahasenack | wireguard-dkms (>= 0.0.20200121-2) | wireguard-modules (>= 0.0.20191219), | 18:08 |
ahasenack | I'm wondering about this in the wireguard MIR (https://bugs.launchpad.net/ubuntu/+source/wireguard/+bug/1950317) if I should seed "wireguard" or "wireguard-tools" | 18:09 |
ubottu | Launchpad bug 1950317 in wireguard (Ubuntu) "[MIR] Wireguard" [Critical, Fix Committed] | 18:09 |
ahasenack | (the depends paste above is missing a line which pulls in "wireguard-tools") | 18:09 |
ahasenack | wireguard-dkms (>= 0.0.20200121-2) | wireguard-modules (>= 0.0.20191219), | 18:09 |
ahasenack | wireguard-tools (>= ${source:Version}), | 18:09 |
ahasenack | I'm wondering about the wireguard-dkms bit, I don't want that in main | 18:09 |
ahasenack | rbasak: do you know? ^ | 18:10 |
rbasak | ahasenack: I think wireguard is just a contentless metapackage? | 18:12 |
rbasak | So -tools only we need | 18:12 |
ahasenack | yes, wireguard is a metapackage | 18:13 |
ahasenack | but so if someone disabled universe, "apt install wireguard" won't work, that's the experience I wanted to avoid | 18:13 |
rbasak | I don't think anyone actually disables universe nowadays | 18:14 |
rbasak | So I'm not sure it's worth caring about that too much. The right thing to do would be to install the wireguard-tools package only. | 18:14 |
ahasenack | ok, agreed | 18:15 |
ahasenack | rbasak: https://code.launchpad.net/~ahasenack/ubuntu-seeds/+git/platform/+merge/415920 pretty please? :) | 18:20 |
jpeisach_ | hey, i'm ItzSwirlz on another account | 18:25 |
jpeisach_ | who maintains libsdl2-dev? it is broken on jammy | 18:25 |
rbasak | ahasenack: done | 18:25 |
jpeisach_ | causes systemd issues, nearly broke my system trying to develop InfiniTime stuff | 18:25 |
ahasenack | rbasak: \o/ | 18:25 |
ahasenack | that seems to be... server team (!) | 18:26 |
ahasenack | https://launchpad.net/ubuntu/+source/libsdl2 | 18:26 |
jpeisach_ | that would explain why it tried going for my desktop components only | 18:26 |
jpeisach_ | im updating my system but if i confirm it i'll dm them | 18:26 |
ahasenack | see if the issue is here, if not, report it: https://bugs.launchpad.net/ubuntu/+source/libsdl2/+bugs?orderby=-id&start=0 | 18:27 |
ahasenack | although, keep in mind there are many transitions going on right now, it might be a temporary issue | 18:27 |
=== lifeless_ is now known as lifeless | ||
=== sem2peie- is now known as sem2peie | ||
=== sem2peie- is now known as sem2peie | ||
rbasak | bryyce: did you know about shuf(1) from coreutils? | 19:22 |
bryyce | rbasak, no, and not surprised I reinvented the wheel with shuffle ;-) | 19:29 |
bryyce | I actually needed that code for shuffling m3u files which needs shuffling pairs of lines | 19:29 |
sarnold | rbasak: re disabled universe, I think the 'live system' installer is main only, every now and then we have to instruct folks how to enable universe to install whatever it is they want installed | 19:54 |
rbasak | sarnold: that's good to know, thanks. | 19:56 |
rbasak | I don't think that changes my opinion though. Those users should `apt install wireguard-tools` if they want it, and `apt install wireguard` would be wrong (since they don't need the DKMS package). | 19:56 |
rbasak | Unfortunately the name of deb packages conflate implementation with UX in users' eyes, so this isn't perfect, but that's how it is. | 19:57 |
rbasak | Or, I suppose we could maintain a delta to drop the wireguard-dkms dependency in Ubuntu since Ubuntu-shipped kernels already include that functionality. But that's maybe not worth it. | 19:58 |
sarnold | yeah, I'm not sure maintaining a delta just to handle this is worth it, but it'd probably be a low-effort delta.. | 19:59 |
seb128 | sarnold, if the debian maintainer is friendly you can get a dynamic depends generation from debian/rules in Debian and stay in sync | 20:15 |
ahasenack | that delta is easy | 20:17 |
ahasenack | I'm just unsure we won't be reintroducing a bug | 20:17 |
ahasenack | the history of that depends ordering is complicated, from what I could gather | 20:17 |
ahasenack | I can ask apw again tomorrow (it's past his EOD) | 20:20 |
mmikowski | Hey team, we at Kubuntu Focus are looking at ways to provide customers with better system rollback solutions. | 20:37 |
mmikowski | We have a working prototype with luks+lvm+ext4 which is nice, but the snapshot sizes grow very quickly and it requires a good deal of additional disk space. | 20:37 |
mmikowski | The other approach we were looking at is luks+btrfs (no luks). This appears to work well as POC, but I was curious what others felt about this setup. | 20:38 |
mmikowski | fwiw, almost-FDE is completely a requirement here. Also, if there is interest in adding an upstream option we can certainly contribute our work. | 20:39 |
ahasenack | mmikowski: ubuntu has zsys, introduced in one of the previous releases, which uses zfs, and a very specific zfs dataset layout | 20:41 |
ahasenack | it integrates with grub, and has the rollback functionality | 20:41 |
ahasenack | I haven't tried it recently, though | 20:41 |
ahasenack | mmikowski: https://github.com/ubuntu/zsys | 20:42 |
ahasenack | and https://discourse.ubuntu.com/t/zfs-focus-on-ubuntu-20-04-lts-blog-posts/16355 | 20:42 |
mmikowski | I really like the idea of ZFS a lot, but a risk analysis looks rough :P | 20:43 |
ahasenack | I think apt has a hook for btrfs, or had at least | 20:44 |
ahasenack | have you looked at apt-btrfs-snapshot ? | 20:44 |
mmikowski | Yes, apt-btfs-snapshot is there. | 20:44 |
mmikowski | Already using it? | 20:44 |
ahasenack | I don't, I'm in the zfs camp ;) | 20:45 |
mmikowski | also snapper | 20:45 |
ahasenack | but I don't use rollbacks in zfs either, I use it just so it's easy to send backups | 20:45 |
ahasenack | regarding your size issue with lvm snapshots, are you doing pruning? Or else you will have the same problem in all solutions | 20:45 |
mmikowski | The alure is to let users to update after a botched upgrade or ill-advised package install. | 20:46 |
mmikowski | *update = rollback | 20:46 |
ahasenack | in the zsys approach, there is a separation between user data and apps | 20:49 |
ahasenack | so rolling back won't rollback changes done to your home, for example | 20:49 |
ahasenack | iirc | 20:49 |
mmikowski | ahasenack: with btrfs, there is also a separation between /home and / on subvolumes. | 20:56 |
ahasenack | note I said zsys | 20:56 |
mmikowski | Rollbacks are nice, and pretty performant on single disk system. | 20:56 |
ahasenack | it's what makes the zsys layout so "complicated", it's not just /home | 20:56 |
mmikowski | ahasenack: is zsys distinct from zfs? | 20:57 |
ahasenack | it's a tool that manages the snapshots and rollback | 20:57 |
ahasenack | using zfs primitives | 20:58 |
ahasenack | and it integrates into the ubuntu desktop | 20:58 |
ahasenack | so you get grub with support for it (you select in the grub menu what previous snapshot you want to boot into), | 20:58 |
ahasenack | adduser creates a new zfs dataset for the user | 20:58 |
ahasenack | and so on | 20:58 |
mmikowski | Thanks. Yes, we see most of this with btrfs. So with a single or dual disk system it seems to provide similar benefits and stability. | 21:00 |
ahasenack | yes | 21:00 |
mmikowski | I do know that zfs is far superior with RAID and enterprise features like those previously available with netapp. | 21:00 |
ahasenack | the debate goes on :) | 21:00 |
mmikowski | But we didn't go down the zfs route too far. | 21:01 |
ahasenack | I personally found it much easier to use, it's a nice cmdline interface | 21:01 |
ahasenack | but there are some messy things | 21:01 |
mmikowski | ahasenack: We are very interested in ZFS and sys. We would need to incorporate somehow into a custom installer. Thank you for the links, and please share if there is more we should do. | 23:26 |
ahasenack | np | 23:26 |
mmikowski | ahasenack: or any other links you recommend :) | 23:26 |
ahasenack | one of those I think has a collection | 23:27 |
ahasenack | of links, I mean | 23:27 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!