[05:11] <chmod> I want to upgrade my server from 18.04 to 20.04 but the only 3rd party repositories I ahve are ondrej repository for php7 and weechat.
[05:11] <chmod> What should I do in this case before I run do release upgrade
[05:25] <quadrathoch2> recommended is removing (ppa-purge) and then install it again after upgrade chmod
[05:25] <chmod> I see.
[07:01] <technoob> Guys
[07:01] <technoob> How do you activate wifi hotspot on an ubuntu server
[07:07] <quadrathoch2> technoob the server should be a hotspot?
[07:07] <technoob> Yes
[07:09] <technoob> quadrathoch2:
[07:09] <quadrathoch2> technoob https://help.ubuntu.com/community/WifiDocs/WirelessAccessPoint
[07:10] <technoob> Where do i start on that
[07:12] <technoob> quadrathoch2:
[07:13] <quadrathoch2> probably with point 3
[07:15] <technoob> quadrathoch2: this section "Setup hostapd, port forwarding, haveged, and dnsmasq"
[07:15] <technoob> ?
[07:15] <quadrathoch2> https://help.ubuntu.com/community/WifiDocs/WirelessAccessPoint#Setting_up_a_bridged_wireless-to-ethernet_router
[07:18] <technoob> Hmm do i need a switch?
[07:19] <technoob> I just want to be able to hotspot without internet
[07:20] <technoob> quadrathoch2:
[07:41] <technoob> Hi
[08:00] <alkisg> Hi, it is possible to install ubuntu-server.iso and completely purge snap afterwards? Or is snap now a requirement for ubuntu server installations?
[08:01] <quadrathoch2> alkisg you can just purge it
[08:01] <alkisg> Thank you quadrathoch2
[08:09] <lordievader> Good morning
[08:11] <quadrathoch2> mornin'
[09:06] <eoli3n> mwhudson TJ- -> vagrant destroy -f && git checkout dev && vagrant up
[09:12] <mwhudson> well yes python is segfaulting
[09:12] <mwhudson> that definitely shouldn't be happening
[09:12] <mwhudson> but also i shouldn't be working this late
[09:14] <TJ-> eoli3n: yeah, looking at it. (Got other important work of my own to do here too)
[09:15] <TJ-> mwhudson: we'll add anything useful to the bug report
[09:15] <TJ-> I'm on GMT/UTC here
[09:28] <eoli3n> no problem
[09:29] <eoli3n> please both confirm me that client IF names are 'ens5' and 'ens6' ?
[09:29] <mwhudson> i have reproduced the failure
[09:30] <mwhudson> and it is very strange
[09:30] <eoli3n> mwhudson sleep on it ;)
[09:30] <mwhudson> i intend to :)
[09:31] <eoli3n> can you just please confirm me IF name, if that differs from what i get here, you will be stuck in another issue
[09:31] <eoli3n> just want to be sure about this
[09:31] <eoli3n> tty2 on client -> ip a
[09:39] <TJ-> eoli3n: strangely, server and client started but whilst I wasn't watching both guests have gone into 'paused' state
[09:39] <eoli3n> TJ- you miss ram
[09:40] <TJ-> eoli3n: I what?
[09:40] <eoli3n> that happened to me when i was reaching maximum RAM
[09:40] <eoli3n> free -h
[09:40] <TJ-> eoli3n: 32GB !
[09:40] <eoli3n> hm strange
[09:40] <eoli3n> oh no
[09:41] <eoli3n> df -h
[09:41] <eoli3n> that your storage
[09:41] <eoli3n> thats
[09:41] <eoli3n> does it ?
[09:41] <eoli3n> no more disk space
[09:41] <mwhudson> ens5 and ens6?
[09:41] <eoli3n> thanks mwhudson that's ok
[09:43] <eoli3n> TJ- please check if /var/lib/libvirt is mounted on a specific storage space and not on your 'too little' slash :)
[09:43] <mwhudson> argh keyboard layout rage :)
[09:43] <eoli3n> yes
[09:43] <eoli3n> comment layout in vagrant file
[09:43] <eoli3n> https://github.com/eoli3n/vagrant-pxe/blob/dev/Vagrantfile#L46
[09:43] <eoli3n> comment that line then, vagrant reload client
[09:44] <eoli3n> i had the same rage when not finding why loadkeys fr didn't work
[09:44] <eoli3n> vnc is forcing layout
[09:45] <eoli3n> TJ- to manage this, you can tweak it by recreating "default" pool from virt-manager
[09:45] <TJ-> eoli3n: yep, /var/ was at 100%
[09:45] <eoli3n> so go to virt-manager
[09:45] <eoli3n> right clic on your connection
[09:45] <TJ-> eoli3n: all fixed now
[09:45] <eoli3n> ok
[09:46] <TJ-> eoli3n: subiquity running
[09:47] <TJ-> eoli3n: apt-config stage reached
[09:47] <TJ-> eoli3n: crash!
[09:47] <eoli3n> :)
[09:47] <eoli3n> now we all use the same env, we can work efficiently
[09:47] <eoli3n> don't forget to comment fr layout in Vagrantfile TJ-
[09:47] <eoli3n> to be able to debug from virt-manager
[09:49] <TJ-> eoli3n: I know where the keys have mapped to :) / -> ;
[09:49] <eoli3n> loadkeys will not work
[09:49] <TJ-> eoli3n: looking at /var/log/installer/subiquity-debug lo
[09:50] <eoli3n> yep nothing more than in the crash file normally
[09:50] <eoli3n> does the installer endless loops ?
[10:31] <TJ-> eoli3n: it did 3 times until I stopped it ("sudo snap stop subiquity")
[10:31] <TJ-> eoli3n: I think I've located the issue; testing it now
[10:33] <mwhudson> it looks like apt is symlinking the package file into /var/lib/apt/partial and then subsequently failing to open it with EACCESS
[10:34] <mwhudson> let me guess overlayfs over nfs is a cauldron of bugs
[10:34] <TJ-> nope :)
[10:35] <TJ-> just waiting for the install to finish to be sure I'm correct
[10:36] <mwhudson> or maybe it is root vs _apt
[10:38] <TJ-> mwhudson: so far, it looks like what I told eoli3n yesterday early on - incorrect file-system permissions on the server (or rather, only allowing root user access) so _apt cannot traverse or read the NFS.
[10:38] <TJ-> mwhudson: the installer is at the "downloading and installing security updates" stage right now
[10:39] <mwhudson> ah
[10:39] <mwhudson> i would also like to know why this leads the installer to segfault
[10:40] <TJ-> mwhudson: all I did, on server, was "find /exports/focal/dists /export/focal/pool -type d -execdir chmod go+x {} \;"  and "find /exports/focal/dists /export/focal/pool -type f -execdir chmod go+r {} \;"
[10:40] <TJ-> mwhudson: the segfaults look to be on the 2nd/3rd runs when /target/ and friends are already mounted
[10:41] <TJ-> mwhudson: it just rebooted after install. I've disabled the pxe libvirt network adapter in the guest, and am trying to aboot
[10:42] <mwhudson> i'm attached to the installer in gdb now
[10:42] <TJ-> guest is booting the installed OS now from its local storage
[10:43] <mwhudson> hmm that didn't work
[10:43] <TJ-> eoli3n: So, it looks like as I said yesterday, you had incorrect file-system permissions on the NFS export
[10:43]  * mwhudson really stops for the night
[10:43] <TJ-> mwhudson: I'm guessing that as apt uses the _apt user when it can, that broke it
[10:44] <mwhudson> TJ-: yes, that side of things makes sense
[10:44] <mwhudson> i wonder if we should stop apt doing that during install
[10:44] <mwhudson> certainly the failure is pretty opaque
[10:44] <TJ-> mwhudson: possibly it uses UID 0 to get the Releases{,.gpg} then drops to _apt for the rest
[10:44] <mwhudson> juliank is the man for that sort of question
[10:44] <TJ-> Guest is fully booted and ran cloud-config etc
[11:30] <frickler> icey: coreycb: jamespage: please see https://bugs.launchpad.net/bugs/1903531 regarding neutron on stein+train, not sure how to best handle this for UCA
[11:30] <icey> frickler: saw the email around that :-/
[11:36] <icey> frickler: is the version currently in the UCA in the middle of that (ie: is impacted)?
[11:39] <frickler> icey: for stein that's 14.4.0 and that is affected, like 14.3.1. seems not many deployments are using ipset, so it took a long time to discover
[11:39] <icey> ah
[11:40] <frickler> or maybe not many deployments updating regularly. for our prod we still have 14.3.0, so would prefer to see a 14.5.0 with the revert in, but of course others may have a different view
[11:44] <icey> frickler: so I fully understand, the versions of neutron that are currently in for stein&train are affected, and the situation is unchanged with the proposed update for stein (https://bugs.launchpad.net/cloud-archive/+bug/1904563)?
[11:48] <frickler> 14.4.0 and 14.4.1 should be compatible, yes
[12:07] <icey> frickler: it sounds like we're ok pushing out the 14.4.1 that's currently in proposed for Stein, and then picking up a new stable point release to get 14.5.0 as soon as the release cuts?
[12:11] <frickler> icey: that would be one option. the other would be to wait for 14.5.0 or include the revert in 14.4.1 already, allowing for an easier upgrade for deployments that are still on <=14.3.0
[12:12] <icey> frickler: how long would you estimate until the 14.5.0 is cut?
[12:12] <frickler> there's also the question how you could notify deployers of the incompatibility, like generate a warning in the pre-install script
[12:13] <frickler> neutron folks are trying to get it out as fast as possible, but there's some issue with the CI it seems. hopefully not more than a couple of days anyway
[13:07] <eoli3n> thanks mwhudson and TJ-, i will test it in few hours
[14:09] <shecki> hi i'm setting up an autoinstall for ubuntu-server 20.04 and am wondering how to set hostname to the DHCP/DNS information. If the installer is run manually, it lists the value I want, but with autoinstall, it comes up as ubuntu-server. Setting it to a value only works for one server but not for a couple servers with different names. is there a way to do this?
[14:42] <eoli3n> shecki "to" DHCP or "from"
[14:42] <eoli3n> ?
[14:46] <eoli3n> https://serverfault.com/questions/1024584/automated-ubuntu-20-04-server-install-how-can-i-get-hostname-from-dhcp-server
[14:48] <eoli3n> https://askubuntu.com/a/1177790
[14:48] <eoli3n> try to set it to "localhost" it well get it from dhcp
[14:50] <shecki> eoli3n: from DHCP. the variant with setting it to localhost will still have localhost as static hostname and only show the given one. is that truly not possible? netcfg/get_hostname= would be the old way and this is quite a must have
[14:51] <eoli3n> open an issue on launchpad
[14:55] <icey> frickler: it looks like the same change is in Ussuri and Victoria, any idea why the change isn't being reverted there?
[15:20] <eoli3n> TJ- your fix doesnt work here
[16:16] <TJ-> eoli3n: really?
[16:29] <eoli3n> really
[16:29] <eoli3n> tried 3 times
[16:29] <eoli3n> with two differents ways
[16:29] <eoli3n> to be sure
[16:29] <eoli3n> maybe you did something more that you forgot ?
[16:31] <eoli3n> TJ-: vagrant destroy -f && vagrant up && vagrant ssh server -c "sudo find /exports/focal/dists /exports/focal/pool -type d -execdir chmod go+x {} \; && sudo find /exports/focal/dists /exports/focal/pool -type f -execdir chmod go+r {} \;" && vagrant up client
[16:34] <eoli3n> but i didn't check if this is a different error to be honest, what i can say is that it fails on the same task
[16:34] <eoli3n> apt-config thing
[16:41] <eoli3n> lets recheck if this is the same error
[16:49] <eoli3n> TJ- same problem
[16:50] <eoli3n> when i chroot /target bash
[16:50] <eoli3n> and apt update
[16:50] <eoli3n> same apt_ error
[16:50] <eoli3n> https://x0.at/Xwi.png
[16:52] <eoli3n> vagrant ssh server -c "sudo ls -altr /exports/focal" : https://x0.at/fEO.txt
[16:53] <eoli3n> -rw-r--r-- 1 root root 18748 Jul 31 17:35 Packages.gz
[16:54] <eoli3n> anyway, i use no_root_squash
[16:54] <eoli3n> so any user should be root on the export
[16:54] <eoli3n> if you made it work, it should be something else, i think
[16:57] <lotuspsychje> eoli3n: failed to fetch, repo issues perhaps?
[16:58] <eoli3n> no repo here, that file is local, served by nfs
[16:58] <eoli3n> no distant* repo
[16:58] <eoli3n> https://bugs.launchpad.net/subiquity/+bug/1905296
[16:58] <lotuspsychje> ah
[16:59] <lotuspsychje> lemme dig some subiquity recent bugs
[16:59] <eoli3n> hm no_root_squash maps only root perms
[16:59] <eoli3n> lotuspsychje i'm talking with the subiquity dev
[16:59] <eoli3n> so he's on it
[17:00] <lotuspsychje> kk
[17:00] <eoli3n> but sleeping :)
[17:20] <lotuspsychje> fresh from the press, not sure if relevant bug #1905599
[17:25] <Odd_Bloke> That's not a direct response to any subiquity issues, just our regular(ish) SRU of cloud-init into stable releases.
[17:25] <lotuspsychje> kk
[17:26] <lotuspsychje> i was watching #ubuntu-bugs-announce but nothing other related recently
[17:29] <Odd_Bloke> :)
[20:04] <DammitJim> when joining an ubuntu 18 server to a windows domain
[20:04] <DammitJim> what is the parameter I need to add to my configs so one doesn't have to specify the domain for a user when logging on?
[20:04] <DammitJim> or maybe you guys can help me with keywords to google this?