[02:04] hello, has anyone got ssl worked with MAAS 2.4? [02:18] partlycloudy: you should try #maas [02:18] partlycloudy: also, learn to ask for help, not for a survey [02:19] leftyfb: thank you leftyfb. i tried #maas, but it was silent there… [02:19] partlycloudy: you'll need to wait, possibly till Tuesday [02:22] leftyfb: oh yeah… long weekend. [07:28] Good morning [11:00] good morning [11:02] when calling dpkg-buildpackage, I get the following error: [11:02] dpkg-source: error: aborting due to unexpected upstream changes, see /tmp/dovecot_2.3.4-2.diff.iemGlV [11:02] What does that mean? My source tree is clean [11:04] it's not, when compared to the orig tarball [11:04] not that sometimes (most of the time, in my experience), the "dh clean" step doesn't really clean it up completely [11:05] How do I fix it? [11:05] unapply all patches, and compare your tree with the orig tarball [11:05] maybe you committed something to the tree instead of as a debian patch [11:06] or maybe there is "dirt" that dh clean didn't clean, so try "git clean -f -x -d" [11:06] and check git status [11:06] with --ignored [12:01] rbasak: what's the difference between "platform" and "ubuntu" in these seeds repositories? https://code.launchpad.net/~ubuntu-core-dev/ubuntu-seeds/+git/ [12:05] ahasenack: I believe it's legacy from when we had separate 3 year and 5 year LTS periods between desktop and server. [12:06] ahasenack: now they're equivalent but just have different seeds in different places. [12:11] rbasak: so how did you find out where to add the irc server seed, anope? [12:20] I'm getting a strange error when I try to dput a ppa: [12:20] gpg: ../dovecot_2.3.4-2_source.changes: error 58: gpgme_op_verify [12:20] gpgme_op_verify: GPGME: No data [12:20] Has anyone seen this before? [12:27] ahasenack: I looked for the seed that contained supported server packages (that aren't seeded anywhere further up the hierarchy) [12:28] kstenerud: did you sign the changes file? [12:44] Hey all. In Ubuntu Server 18.04 docker is a "snap". How do I do "... --volume /path:/mount ..." in such environment? [12:45] I need to specify a volume to be in a dedicated disk/partition mounted in the host. [13:18] rbasak: hm, last time I did a seed change, it was in "ubuntu" repository [13:18] but my change was to drop a seed [13:18] so I just had to find where it was defined [13:18] but it's still unclear, I just asked in #ubuntu-devel [13:50] rbasak: did you see the response in #ubuntu-devel about the difference between "ubuntu" and "platform"? [13:58] ... [15:00] it is a plausible theory that someoene who had the ubuntu-server package but no -desktop packages installed on 16.04, but had xfce and xserver-xorg installed, and initiated a do-release-upgrade upgrade to 18.04, would end up with xwayland? [15:09] we also dont know what the user has chosen during upgrade, he did mention to have chosen yes-no etc [15:20] tomreyn: I think you're better off asking desktop people. Server people don't deal with xfce, xserver-xorg or xwayland. [15:20] Server packages have no interaction with those components. [15:22] rbasak: and this is entirely logical. but i assumed server folks do design the upgrade path for when there is the ubuntu-server package installed, and i assume this was the case here. [15:23] but let's not spend time on it, since i have no hard facts. [15:23] thanks [15:32] tomreyn: understood. FWIW, there isn't an upgrade path as such. Upgrade paths are implemented when necessary in individual packages. [15:40] oh, thanks for pointing this out, so that's a misconception i had there. i thought i had noted that the release upgrader code which is downloaded by do-release-upgrade inspects presence of these meta packages and makes choices based on those findings. [15:43] is there any crontab that can support year in crontab? I just want to run job only once (without deleteing crontab entry for example) I can not use at, because I need some shared "db". I am able to create lock file with flock to prevent duplicate triggering. [15:45] muhaha: how about a little quirk like: [ "$(date +%Y)" = "2020" ] && myjob [15:49] tomreyn: do-release-upgrader does have some magic for some edge cases I believe, but that's the exception rather than the norm. [15:49] It's not that the server team writes an "upgrade path" for each release or anything. [15:49] muhaha, what do you mean by a shared db? [15:50] I'm not sure the server team has ever needed anything in do-release-upgrade. [15:50] rbasak: thank, this is is good to know, should prevent me from providing as much flase info in the future :) [15:50] We generally just make sure we ship stuff that'll upgrade correctly with plain apt. [15:52] I wouldn't be surprised to hear that the switch to wayland and back needed some magic in do-release-upgrade. However I don't know either way. Such a thing would have been implemented between the desktop and foundations teams. [15:53] lordcirth_: I have webhook wrapper which can setup crontab for example... I can run this server in failover/more that 1 instance... atd can not use NFS for spool, but crontab depends only on files on FS, so If I create lock when is time to trigger job, another instance can not trigger this job as duplicate, if is there lock file from flock. [15:54] muhaha, ah ok [15:54] Problem is that crontab can not hanlde one time jobs.. [15:54] nncron can use year in pattern AFAIK [15:54] muhaha, if the first thing your job does is check the lock file, it can just do nothing [15:55] no, one of X instances will trigger crontab job-> first one will create lock file and run command, otheres will not run because of lock file -> already created [16:01] muhaha, yeah, that's what I mean. Just don't delete the lock file when you're done. Then it will only run once. [16:01] http://www.nncron.ru/help/EN/working/cron-format.htm -> this one can handle year -> you can trigger job only once with shared file locking [16:04] lordcirth_ I am using https://linux.die.net/man/1/flock -> can not create and let lock file forever, some bash wrapper is wacky ... I just want quick, proper way. [16:04] Well, nncron seems like it works too, as long as it's not hard for you to migrate [16:07] muhaha: which release are you using? Sounds like a job for a systemd timer [16:13] Oh, yeah that exists too [16:16] systemd time is not in dockerized ubuntu, right? [16:44] muhaha, probably not. The host sets the kernel time. === TheHonorableKitt is now known as THKitten [16:46] is there anyone in here who is an admin in #linux? Looks like I've been banned again (because I'm using TOR). This was fixed before for me as I'm a trusted user, but it looks like it's been broken. [17:30] THKitten: did you try in ##linux-ops ? [17:32] kstenerud: schleuder is also failing in debian at least: https://ci.debian.net/packages/s/schleuder/ [17:32] same failure [17:33] ruby... === logan_ is now known as logan- [18:55] i upgraded from 14 to 16, and it broke booting. it seems to be related to lvm and initrd/initramfs [18:56] grub loads ok, and the kernel starts to load, but then when it tries to mount various filesystems, it can't, and eventually gives up trying and drops to an initramfs prompt [18:57] all i have to do to get it to work is manually issue "vgchange -y", then i can exit and it resumes booting and everything is fine - so it seems like the initrd has support for lvm, but just isn't activating the volume groups? [18:58] how can i troubleshoot and fix this? [18:58] i've since upgraded from 16 to 18.04, and 18.04 to 18.10, but it remains broken [18:59] lunaphyte, https://bugs.launchpad.net/ubuntu/+source/lvm2/+bug/1573982 [18:59] Launchpad bug 1573982 in lvm2 (Ubuntu) "LVM boot problem - volumes not activated after upgrade to Xenial" [Undecided,Confirmed] [19:01] lordcirth_: oh, wow. was that something you happened to already know about, or did you just now find it with some searching? [19:01] FWIW, this sort of thing is why I just reinstalled [19:01] lunaphyte, I searched it just now [19:01] grr [19:01] "ubuntu initrd lvm" [19:02] well, i'm grateful you did, for whatever reason, i was obviously doing an awful job [19:02] lol np [19:02] lordcirth_: why does reinstalling make the problem a non issue? [19:02] lunaphyte, pretty sure the 16.04 installer works fine with LVM. It's upgrading the old system that misses something. [19:03] I'd seriously consider wiping it and installing 18.04 [19:03] i see [19:03] yeah, i would like to, honestly [19:03] not right now though [19:03] i'll be able to once i get some other crap straightened out [19:04] lunaphyte, also, just wondering, do you need LVM? I don't use it anymore [19:04] i guess that depends on what "need" means :) [19:04] I just reinstalled my home PC with ZFS root, which was a bit involved, but it's awesome now [19:04] do i need it? no, not really. i do like it though [19:04] oh zfs [19:05] yeah, maybe at some point [19:05] lunaphyte: haha! Notice I posted a udev patch to that bug? [19:05] btrfs works as well and is trivial to install to /, unlike ZFS [19:06] TJ-: oh, let me look [19:06] But I wanted raidz / raid5, and btrfs's raid5 is alpha [19:06] lordcirth_: i have been using btrfs on other systems recently [19:06] i do like it, but i've unfortunately had some very very bad experiences too [19:06] It's a bit buggy still. ZFS is wonderful. [19:06] i'm hoping i've just done something irresponsible, but that's another thing to figure out [19:07] systems that get shutdown uncleanly, and the filesystem is rendered completely unusable [19:07] repeatedly [19:07] it was a huge bummer [19:07] but anyway, that's a different story [19:08] TJ-: how do i see the patches posted to the bug? just within the thread discussion itself? [19:09] lunaphyte: Looks like my patch addresses a slightly different case to yours; in my case it was when kernel command line has root=UUID= whereas yours is for /usr not being activated [19:09] oh ok [19:09] lunaphyte: yes, they're linked from the comment where they are introduced, and aslo listed on the right side [19:09] oh there, i see. thanks [19:10] lunaphyte: Your /usr/ issue won't be solved by that, you need the earlier patch form comment #10 I think it is [19:11] "I then booted via rescue system and added "lvm vgchange -ay" in /usr/share/initramfs-tools/scripts/local-top/lvm2 right before "exit 0"" [19:11] that guy? [19:11] Correct [19:11] ok [19:11] use -aay rather than -ay though [19:11] is that a more so a hack? or is it actually the correct way to solve the problem? [19:14] lunaphyte: correct way - the problem occurs because this command was removed from the software [19:14] ok, cool. thanks very much [19:14] i'll give it a try shortly [19:15] don't forget to update-initramfs beforehand [19:15] before rebooting? [19:19] rbasak: what I meant in standup, I can't install libmariadbclient-dev in debian/sid: https://pastebin.ubuntu.com/p/7bkNKWdGbf/ [19:20] dev one is 10.1, whereas the one carrying the soname is 10.3 [19:38] lunaphyte: yes, after adding change and before rebooting [19:38] gotcha. i won't forget :p [20:04] in /usr/share/initramfs-tools/scripts/local-top/lvm2, i see "activate "$ROOT"", which makes sense, but also "activate "$resume"" [20:05] i don't see where $resume is getting set. what is this? where does it come from? [20:05] How do you install lzma or lz4 support for mariadb in this OS? [ERROR] InnoDB: innodb_compression_algorithm = 4 unsupported. InnoDB: liblzma is not installed. Yet apt cache certainly thinks liblzma5 is installed. Does ubuntu not support anything but gzip for row_format compression? [20:07] version 10.1 [20:25] Its, cool. No one uses lz4 for realtime compression anyway due to libz superior compression speed and ratio. [20:31] i'm having trouble getting this system to keep a resolution of 1280x1024 [20:32] i've set GRUB_GFXMODE=1280x1024 and GRUB_GFXPAYLOAD_LINUX=keep in /etc/default/grub, which seems to partially work, but something is still changing the resolution at some point during boot [20:32] the display i'm using can't go beyond that, so it goes "out of range" when that happens and i can't see the screen of course [20:32] how can i figure out what is doing this? [20:50] it looks like maybe due to the edid information? [20:57] aha, yes [20:58] booting with the physical display worked. the kernel didn't detect it, so it left the resolution alone, and the network console didn't go out of range [21:00] oops [21:00] "booting with the physical display disconnected worked", i meant to say [21:04] ahasenack: libmariadbclient-dev is deprecated in 10.3 I think and will be removed. [21:23] Are our regular cloud images supposed to work on bare metal? What does MAAS do? [21:24] I have a friend for whom this doesn't work because his USB keyboard doesn't work. Installing linux-image-generic fixes it. [21:24] He can't use MAAS because he needs the installation to work over the Internet. [21:24] So no TFTP. [21:25] In his case a USB installation is fine, so he's blatting the cloud image onto a USB stick at my suggestion, and separately providing a NoCloud cloud-init datasource using an SD card. [21:25] (well TFTP would work, but he can't control DHCP to bootstrap that) [21:26] besides https://docs.maas.io/2.5/en/installconfig-network-ssl your maas service need a lot of work for ssl https://askubuntu.com/questions/736126/using-https-with-maas-web-interface-login-redirects-to-http [21:27] This approach seems to work fine except he's currently needing to make two modifications to the image that he'd like to avoid: 1) installing linux-image-generic to make the USB keyboard work; and 2) dropping console=ttyS0 to work around bug 1573095 (I'm about to update that bug with details) [21:27] bug 1573095 in Ubuntu "16.04 cloud image hangs at first boot" [Undecided,Confirmed] https://launchpad.net/bugs/1573095 [21:27] rharper: ^ would you happen to know please?