[11:15] <cpaelzer> sil2100: doko: vorlon: rcj:  fginther: tobikoch: I was just made aware (thanks tomreyn) that http://releases.ubuntu.com/18.04/ubuntu-18.04.2.0-live-server-amd64.iso.torrent on https://www.ubuntu.com/download/alternative-downloads#bittorrents is a 404 - I'd expect it to point to http://releases.ubuntu.com/18.04/ubuntu-18.04.2-live-server-amd64.iso.torrent instead
[11:16] <cpaelzer> uh spam alert, but I just didn#t know who would pick tihs one up :-)
[11:17] <rbasak> That might be one for IS (so #canonical-sysadmin).
[11:19] <juliank> rbasak: i think it's design and the bug is an extra .0 in this page: https://github.com/canonical-websites/www.ubuntu.com/blob/master/templates/download/alternative-downloads.html
[11:23] <infinity> (For the record, the above is being sorted)
[11:35] <tomreyn> (And for the record, the real credit goes to 'yn' in #ubuntu, I just forwarded.)
[12:26] <seb128> cpaelzer, on bug #1815978 your comment describe bionic details but the report/title is about xenial?
[12:38] <seb128> hum
[12:38] <seb128> https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-disco/disco/armhf/s/software-properties/20190215_033903_76f69@/log.gz
[12:38] <seb128> failed on
[12:38] <seb128> 'Cannot add PPA: 'ppa:~xnox/ubuntu/nonvirt'.
[12:38] <seb128> The user named '~xnox' has no PPA named 'ubuntu/nonvirt''
[12:39] <seb128> xnox, ^ is that normal it uses your personal ppa?
[13:46] <kstenerud> Does anyone here know how I could make a service that can, when an env is set, prevent the service from actually starting, but signal success so that the installer doesn't bail out?
[13:46] <kstenerud> Or something similar... I need to update an initd script that does exactly that to a systemd script.
[13:47] <rbasak> kstenerud: can you explain more about the problem you're trying to solve? How are you sure that's the right approach to take, for example?
[13:48] <kstenerud> I don't know if I'm taking the right approach. The package needs a configuration, but doesn't provide one, so when the package is installed and started, it fails, and the installer fails.
[13:48] <kstenerud> So the init script for it checks for a $RUN variable, and if set to "no", prints a message saying the service didn't really start, and please make a config file and change the run variable to "yes"
[13:48] <rbasak> What do you mean by "installer"?
[13:48] <kstenerud> apt
[13:49] <rbasak> Which package is it?
[13:49] <kstenerud> freeipmi
[13:49] <kstenerud> it's been broken since cosmic
[13:49] <kstenerud> https://bugs.launchpad.net/ubuntu/+source/freeipmi/+bug/1808637
[13:50] <rbasak> Is it definitely impossible for the package to ship with a configuration that does work?
[13:51] <kstenerud> freeipmi fails because no nodes are defined. And a node is whatever "thing" a user happens to want to make
[13:51] <kstenerud> so maybe I could make some sort of default "thing", but I don't know enough about the package to know if it's possile
[13:55] <rbasak> Is Debian impacted by this?
[13:58] <rbasak> There is ConditionPathExists, but I don't see anything for an environment variable specifically.
[13:58] <kstenerud> yup, just tested on debian and it also fails on postinstall
[13:59] <rbasak> It's also possible to arrange for the service to not be enabled by default and expect the user to enable it, but then that could be different from init.d and therefore confusing.
[13:59] <rbasak> I don't see any way to get an ExecStartPre failure to cause systemd to treat the service as not failed, but perhaps that's possible.
[13:59] <rbasak> So there are a variety of options to solve the problem.
[13:59] <kstenerud> The initd one just spits out a message telling the user to make a config and then change a setting to make the script really load the service
[13:59] <rbasak> Maybe best to see which path the Debian maintainer takes.
[14:00] <kstenerud> debian side is broken too. I just checked a sid container
[14:00] <rbasak> Does a bug exist in Debian, and if not please could you file one?
[14:00] <rbasak> Another way might be to make ExecStartPre cause a hard fail but don't attempt to start the service on install.
[14:00] <kstenerud> so policy-wise, does this mean hold off on merge until debian fixes it?
[14:00] <rbasak> But then it'd show as failed after next boot.
[14:31] <Skuggen> You can set a service to disabled by default, can't you?
[14:31] <Skuggen> We do this with the multi-instance service for MySQL upstream
[14:35] <Skuggen> Oh, you already said that. nm :)
[14:43] <rbasak> kstenerud: sorry, just saw your question. It's our shout policy-wise. The downside of fixing it now is the extra effort and possible UX problems if the Debian maintainer decides to fix it in a different way.
[14:44] <rbasak> So I would give the Debian maintainer a few days to respond at least.
[17:34] <jmgb4> Hey, just a quick question about how ubuntu packages things. Is there a way to see how a specific package (remmina) pulls down stuff like openssl and what the requirements are? Like a make file or an ebuild (like in gentoo)?
[17:42] <teward> jmgb4: what do you mean by 'pulls down stuff'?  Most packages in Ubuntu are built based off the *other* packages and pulls them in at build/run time into the environment for use
[17:42] <jmgb4> Dependencies teward
[17:43] <teward> package dependencies are in debian/control for the given package, you'd have to download the source package and read through its dependencies... or read through the package page - https://packages.ubuntu.com/bionic/remmina
[17:43] <teward> which lists the deps :P
[17:44] <teward> as for 'pulling them in' they're *typically* already packaged in the repos and are 'pulled in' by the underlying build daemons as part of the build env setup process
[17:44] <teward> via apt, for example, pulling in the dependent packages listed for the given packages
[17:44] <teward> unless i'm not understanding your question well :|
[17:54] <xnox> seb128, it was using my ppa because of my utf-8 characters in the launchpad ppa key
[17:54] <xnox> seb128, it's on my list investigate, as to what's wrong with it.
[18:08] <jmgb4> I like it how you have to put pulled in in quotes everytime teward
[18:48] <vorlon> cpaelzer: can you flag that to the web team if it's not already fixed? the link on https://www.ubuntu.com/download/alternative-downloads#bittorrents should be reset to our normal schema (it was changed temporarily because we had to issue an 18.04.1.0)
[19:47] <infinity> vorlon: I escalated it to them and it was fixed moments after it was noticed.
[20:50] <danboid> Does the current disco installer have ZFS root support?
[20:51] <sarnold> I don't believe so
[20:51] <danboid> I wrote a ZFS installer for Arch and I'm hoping that Ubuntu will default to using a config that will work with zedenv boot environments, like ALEZ does
[20:52] <danboid> https://github.com/danboid/ALEZ
[20:53] <danboid> Arch Linux Easy ZFS installer
[20:54] <sarnold> "zedenv is still experimental and should not be used on production systems"
[20:54] <danboid> sarnold, Yes, but it does actually work and its best to plan ahead
[20:57] <danboid> Is Ubiquity still going to be the installer in 19.10+?
[20:58] <danboid> I mean, is it still the installer
[21:00] <danboid> launchpad seems to think so
[21:01] <sarnold> I don't follow the installer very closely, but I think at this point ubiquity will be 19.10 desktop, subiquity 19.10 default server install image, debian-installer for the alternative desktop and alternative server..
[21:06] <jbicha> danboid: there is work being done on a new Ubuntu Desktop installer, it's being mentioned in https://community.ubuntu.com/c/desktop/team-updates
[21:07] <jbicha> I assumed it was more related to https://lists.ubuntu.com/archives/ubuntu-devel/2018-May/040301.html than to subiquity
[21:10] <danboid> I've opened a ticket as a 'reminder' https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1816190
[21:13] <jbicha> danboid: those weekly reports mention that the zfs root thing is an experiment with the new WIP installer
[21:13] <jbicha> the zfs thing may have nothing to do with using zfs on your computer; it might only be about internal structuring of the live images
[21:14] <jbicha> now you know just about as much as I do on the topic :)
[21:14] <sarnold> danboid: nice, thanks :)
[21:15] <danboid> sarnold, NP!
[21:20] <danboid> You can install to ZFS with maas currently but you get zero options, there is no support for mirrored vdevs, RAIDZ etc, and it only supports installing to UEFI. I hope these restrictions don't last long
[21:22] <sarnold> no mirrored vdevs? that's pretty minimal :)