[03:13] <arch-nemesis> I'm attempting to use frr with BGP unnumbered on Ubuntu 20.04. A common way to do this is to set a /32 address to the loopback address, but I can't for the life of me get the address to stay using netplan. Why can't I add additional IP addresses to lo using netplan?
[03:16] <arch-nemesis> Nevermind. I take back my question. It works perfectly, if you don't mis-format your yaml.
[11:23] <mrkewl20> what do you guys recommend for me to use to establish a remote desktop session with my ubuntu server
[14:18] <frickler> jamespage: coreycb: seems we need git-review>=1.27 for the updated opendev gerrit, is there a way to get that backported into bionic/universe?
[14:19] <coreycb> frickler: maybe, unless there's a specific patch to backport, that might be easier
[14:25] <eoli3n_> serioulsy, subiquity is a shame
[14:25] <eoli3n_> what is that output ..? am i a machine ?
[14:26] <eoli3n_> an output needs to be understanble for a HUMAN !
[14:26] <eoli3n_> output is not debug by default
[14:26] <eoli3n_> new tool, but nothing is human usable
[14:27] <eoli3n_> i'm sad, because, it was already a shame that kickstart was not finished for many years. Ubuntu finally use a new tool, good idea, but bad implementation
[14:27] <coreycb> frickler: let's track it here https://bugs.launchpad.net/ubuntu/+source/git-review/+bug/1905282
[14:31] <eoli3n_> https://bugs.launchpad.net/subiquity/+bug/1905178
[14:31] <eoli3n_> what a debug msg :) is BOOTx64.EFI a file system ?
[14:32] <eoli3n_> does lennart poettering works for canonical now ?
[14:33] <eoli3n_> what a joke : https://x0.at/vEE.png
[14:33] <eoli3n_> new tool ! wow, clever work
[14:36] <eoli3n_> installer is looping 3 times configuring partitions and installing system :D
[14:36] <eoli3n_> i don't know if i need to cry or laugh
[14:36] <eoli3n_> maybe both
[14:38] <eoli3n_> endless loop :)
[14:38] <eoli3n_> lets debug that great work
[14:40] <frickler> coreycb: thanks. I guess it would be possible to find a minimal set of patches if needed, but IMO it would be better tu use 1.27 directly, that's the same as in focal, too
[14:45] <coreycb> frickler: ok just looking through the commits now. do you know if there's a workaround?
[14:47] <frickler> coreycb: I don't know, best ask infra-root in #opendev
[15:58] <eoli3n_> ok so
[15:58] <eoli3n_> i want to install ubuntu server with autoinstall
[15:58] <eoli3n_> i don't want to make all my client dl the whole iso because its 900MB
[15:58] <eoli3n_> and i deploy 800 hosts, grouped by 40
[15:58] <eoli3n_> so i made an nfsroot
[15:59] <eoli3n_> installer crash, when i read crash report i see
[15:59] <eoli3n_> E: Failed to fetch file:/cdrom/dists/focal/main/binary-amd64/Packages  File not found - /cdrom/dists/focal/main/binary-amd64/Packages
[15:59] <eoli3n_> http://sprunge.us/ToqZRL
[15:59] <eoli3n_> but i don't understand if crash starts here or not
[16:01] <eoli3n_> crash start at  2020-11-23 15:10:07,224 ERROR root:39 finish: subiquity/InstallProgress/install/curtin_install: FAIL: Command '['systemd-cat', '--level-prefix=false', '--identifier=curtin_log.5282', '/snap/subiquity/1966/usr/bin/python3', '-m', 'curtin', '--showtrace', '-c', '/var/log/installer/subiquity-curtin-install.conf', 'install']' returned non-zero exit status 3.
[16:04] <eoli3n_> my tests are fully reproductible if you use vagrant+libvirt : https://github.com/eoli3n/vagrant-pxe/tree/dev
[16:04] <eoli3n_> the whole config is here
[16:05] <eoli3n_> https://github.com/eoli3n/vagrant-pxe/tree/dev/roles/ubuntu/files
[16:06] <quadrathoch2> eoli3n_ wouldn't it make sense to use the legacy installer?
[16:07] <quadrathoch2> i'm pretty sure by 22.04 autoinstall is aged well enough to use it in production
[16:08] <tomreyn> this should probably not happen: [  147.809312] python3[1872]: segfault at 7f79cfc65438 ip 00007f79cfc65438 sp 00007ffd353b7108 error 15 in libstdc++.so.6.0.25[7f79cfc63000+a000]
[16:09] <eoli3n_> tomreyn is there anything more specific ?
[16:09] <eoli3n_> how do i debug this ?
[16:09] <tomreyn> it's from your logs
[16:09] <tomreyn> i don't see anything more specific there
[16:09] <eoli3n_> that file is the crash file
[16:09] <eoli3n_> so, bug report ?
[16:10] <tomreyn> which installer are you using?
[16:10] <eoli3n_> https://github.com/eoli3n/vagrant-pxe/blob/dev/roles/ubuntu/tasks/main.yml#L4
[16:11] <eoli3n_> is that what you're asking for ?
[16:11] <tomreyn> yes, i was wondering which iso you're using
[16:12] <tomreyn> so file a bug, try getting more infos. this is probably the same thing you previously described as "<eoli3n_> installer is looping 3 times configuring partitions and installing system :D"
[16:12] <tomreyn> how much ram do these vms have?
[16:13] <eoli3n_> https://github.com/eoli3n/vagrant-pxe/blob/dev/Vagrantfile#L37
[16:13] <tomreyn> right, i just spotted it on the dmesg
[16:13] <tomreyn> i assume this can be enough
[16:13] <eoli3n_> 2G is decent
[16:14] <eoli3n_> i can try with more
[16:14] <tomreyn> with zfs, probably a good idea
[16:18] <tomreyn> you should probably also verify the downloaded iso against its checksum at leats, if not against the gpg signature
[16:18] <eoli3n_> ok, i do it just after
[16:18] <eoli3n_> but autoinstall defaults to zfs ?
[16:19] <eoli3n_> i don't think, because i can read ext4 over lvm setup
[16:19] <eoli3n_> in crash file
[16:19] <eoli3n_> "zfs filesystem is not supported in this environment"
[16:19] <tomreyn> oh i didn'T check your specific partition layout, i just noticed zfs modules were loaded
[16:20] <eoli3n_> just wondering, why isn't zfs supported on my env ?
[16:24] <tomreyn> 2020-11-23 15:10:07,224 ERROR root:39 finish: subiquity/InstallProgress/install/curtin_install: FAIL: Command '['systemd-cat', '--level-prefix=false', '--identifier=curtin_log.5282', '/snap/subiquity/1966/usr/bin/python3', '-m', 'curtin', '--showtrace', '-c', '/var/log/installer/subiquity-curtin-install.conf', 'install']' returned non-zero exit status 3.
[16:24] <tomreyn> this and following lines seems to indicate the pyton error
[16:24] <eoli3n_> https://bugs.launchpad.net/subiquity/+bug/1905296
[16:25] <eoli3n_> yes, but it doesn't tell more
[16:26] <eoli3n_> subiquity-info.log -> http://sprunge.us/gEzIZl
[16:28] <eoli3n_> subiquity-curtin-install.conf -> http://sprunge.us/bfyQIa
[16:49] <eoli3n_> https://bugs.launchpad.net/subiquity/+bug/1905298
[16:55] <eoli3n_> quadrathoch2 just see your comments
[16:55] <eoli3n_> missed it
[16:56] <eoli3n_> i don't want to use legacy installer, but i precisly do that work, to be able to install in UEFI
[16:56] <eoli3n_> and i deploy 20.04 because we only deploy LTS versions
[16:56] <eoli3n_> s/but/because
[16:57] <quadrathoch2> eoli3n_ oh yeah I forgot that you want to use uefi :/
[16:57] <eoli3n_> i use netboot (60MB) installer in legacy
[16:57] <eoli3n_> which is great
[17:19] <petrhl> Hi, I'd like to run my custom mysql server on Ubuntu 20.04 with custom directory location. The command is:
[17:20] <petrhl>  mysqld --defaults-file=/home/petr/.config/cqrlog/database/mysql.cnf --datadir=/home/petr/.config/cqrlog/database --socket=/home/petr/.config/cqrlog/sock --port=64000
[17:20] <petrhl> but nothing happened. I tried the same with --initialise command but the same problem. In the error log, there is this: https://pastebin.com/hYQEgbYG
[17:21] <petrhl> I know that it could be blocked by apparmor but I already have @{HOME}/.config/cqrlog/database/  r,
[17:21] <petrhl> @{HOME}/.config/cqrlog/database/** rwk,
[17:21] <petrhl>  in /etc/apparmor.d/local/usr.sbin.mysqld
[17:22] <petrhl> I have no idea why it does not work. There was not any problem in Ubuntu 18.04 but it does not work with 20.04 and newer. Could you help me, please? I already spent many hours googling but without any success :(. Thanks a lot.
[17:26] <petrhl> there is output of --verbose --help added to the end of the command https://pastebin.com/WXpMbvk2
[17:34] <tomreyn> petrhl: do you see, using journalctl, indications that apparmor is blocking the mysqld, though?
[17:36] <petrhl> tomreyn, no, nothing about apparmor or mysqld at all :(
[17:39] <tomreyn> petrhl: nothing about "audit" either?
[17:39] <tomreyn> i.e. run    sudo journalctl -f    then try starting the mysql server
[17:40] <tomreyn> ...and check what was added to the journalctl output
[17:40] <tomreyn> if it's not apparmor related (it would log about it, i would think), then my guess is rather that it's specific to your myswld build
[17:40] <tomreyn> *mysqld
[17:40] <petrhl> no, nothing was added to the journalctl output
[17:42] <petrhl> I use mysqld from Ubuntu repositories
[17:59] <tomreyn> oh, you wrote you're using a "custom mysql server" initally.
[18:01] <tomreyn> i imagine it can be just a matter of permissions (the mysqld process not being allowed all the way into the userS' home directory, or expecting to chroot but not having permission to do so)
[19:40] <teward> which might show in syslog directly instead of journalctl if something isn't reporting right to journal
[19:40] <teward> (apparmor denies for instance sometimes don't report right)
[19:54] <petrhl> teward, You are right! In the syslog, there is udit: type=1400 audit(1606161199.093:8039): apparmor="DENIED" operation="mknod" profile="/usr/sbin/mysqld" name="/home/petr/.config/cqrlog/sock.lock" pid=6838 comm="mysqld" requested_mask="c" denied_mask="c" fsuid=1000 ouid=1000
[19:54] <petrhl> teward, you were right
[19:56] <tomreyn> hmm, interesting.
[20:03] <teward> petrhl: i know i'm right i've had those same problems with customized BIND9 configs ;)
[20:03] <teward> tomreyn: looksl ike they don't have write/change/edit perms in that directory area
[20:04] <teward> 'c' needs allowed i think too to create the socket lock via mknod
[20:07] <sdeziel> petrhl: those should show up in "sudo journalctl -k"
[20:08] <teward> at least we *saw* the errors now xD
[20:14] <petrhl> sdeziel, unfortunately it didn't show anything
[20:15] <sdeziel> petrhl: -k should display the kernel ring buffer... at least it has always worked here
[20:15] <tomreyn> teward: i found it "interesting" that, as you pointed out, sometimes audit logs won't end up in the systemd journal - that's good to know.
[20:16] <sdeziel> tomreyn: are you passing -k to journalctl?
[20:16]  * sdeziel doesn't like problem happening *sometimes*
[20:16] <teward> sdeziel: if we pass -k and aren't seeing audits ast in petrhl's case, then the problem remains as "journalctl and auditd/apparmor not reporting where it should be"
[20:16] <teward> never seen any AA Audit data go into the unit's journalctl
[20:16] <teward> and i had struggled with that for named an eon ago
[20:16] <tomreyn> sdeziel: it didn't happen to me, i just referred to what teward reported
[20:17] <teward> found it in syslog though xD
[20:17] <sdeziel> teward: indeed, those should not go into the unit's log. But those should really make it to the kernel log's (aka dmesg)
[20:17] <teward> they might but might not be ending up in journalctl in the right spot or such.  we could always dig
[20:18] <teward> but at least in petrhl's case we FOUND the audit failure message ;)