[06:06] <lordievader> Good morning
[08:00] <sahid> jamespage: o/ I need to bump version of python-mimeparse that to 1.5.2, can you fork the package in ubuntu-server-dev?
[08:00] <sahid> that is required by python-falcon
[08:00] <sahid> https://launchpad.net/~sahid-ferdjaoui/+archive/ubuntu/eoan-train-proposed/+build/17187763
[08:00] <sahid> https://code.launchpad.net/~sahid-ferdjaoui/ubuntu/+source/python-falcon/+git/python-falcon/+ref/master
[08:20] <munsking> Hello, somewhere on my ubuntu server there's a script or something that tries to log on to our windows domain with the wrong credentials, what's the easiest/fastest way to find our what does that? i could try grepping for the username but that's going to take hours
[08:24] <tomreyn> munsking: sssd maybe? check if you have it installed: dpkg -l sssd
[08:24] <tomreyn> (if the last line of output starts with ii then it's installed.)
[08:25] <tomreyn> https://help.ubuntu.com/lts/serverguide/sssd-ad.html.en
[08:25] <munsking> tomreyn, nope, not installed
[08:27] <munsking> ooh i think i might have found something, could be used for webmin/virtualmin
[08:28] <jc_> Hi, I have installed ubuntu server on my dell server on 370GB SSD, but it is using only 4GB space for path - (/). After installing some applications, I am getting following message "There is not enough disk space on the image storage media." Can anyone suggest how to improve space for / (path).
[08:28] <jc_> # lsblk -o NAME,FSTYPE,SIZE,MOUNTPOINT,LABEL
[08:28] <jc_> sdk                                    372G
[08:28] <jc_> __sdk1                                   1M
[08:28] <tomreyn> munsking: sudo netstat -pan --inet    or   sudo lsof -i    can be used to list open network connections. and here's a list of ports used for AD: https://isc.sans.edu/diary/Cyber+Security+Awareness+Month+-+Day+27+-+Active+Directory+Ports/7468
[08:28] <jc_> __sdk2                    ext4           1G /boot
[08:28] <jc_> __sdk3                    LVM2_member  371G
[08:28] <tomreyn> !paste | jc_
[08:28] <jc_>   __ubuntu--vg-ubuntu--lv ext4           4G /
[08:29] <jc_> okey Thanks. I will use paste.
[08:29] <tomreyn> jc_: not actually needed.
[08:30] <tomreyn> jc_: ubuntu server will only create a single LVM2 logical volume (LV) by default, just large enough to install ubuntu on it. you can either expand this existing LV or add new ones to get more usable file system space.
[08:30] <jamespage> sahid: yep one second
[08:31] <jc_> output of lsblk : https://paste.ubuntu.com/p/tTZ2QK3FPw/
[08:31] <munsking> tomreyn, i think i'd have to look at the exact time of the failed login, which seems a bit random
[08:32] <jamespage> sahid: looks like you need some uploads sponsored?
[08:33] <jc_> tomreyn: Thanks for the info. If I create new LV, all the applications will be installed directly on the newly created LV or do I need to make some changes for that?
[08:34] <tomreyn> jc_: the last line of your lsblk output is the ubuntu OS LV the installer created.. the last but one line shows that you have 371 GB space available to LVM2, so you can create plenty of new LVs or extend existing ones.
[08:35] <tomreyn> jc_: creating an LV is equivalent to creating a partition - you will still need to create a file system on it, and mount it somewhere. and then only data written to this somewhere will end up on this file system.
[08:35] <tomreyn> jc_: e.g. you can create a 50 GB LV, format it as ext4, mount it to /var/lib/mysql and use it for mysql databases.
[08:36] <jc_> tomreyn: I will look in to extending the existing one. I don't want to create new one for atleast this SSD. Thanks
[08:36] <jc_> tomreyn: thanks for your explanation.
[08:37] <tomreyn> jc_: to resize the existing LV and the file system on it to 10 GB (up from 4GB), you should be able to use something similar to:  sudo lvresize -L 10G --resizefs ubuntu-vg/ubuntu-lv
[08:38] <jamespage> sahid: python-mimeparse repo seeded
[08:38] <jamespage> sahid: looking at your other sponsor requests now
[08:40] <jc_> tomreyn: can i just use # lvextend -L+371GB /dev/mapper/ubuntu--vg-ubuntu--lv
[08:50] <tomreyn> jc_: you'll want to also resize the file system, so specify --resizefs. also, i don't think you can specify the path as /dev/mapper/ubuntu--vg-ubuntu--lv but it has to be VG/LV
[08:51] <tomreyn> the lvresize(8) man page states: "lvresize resizes an LV in the same way as lvextend and lvreduce. lvresize resizes an LV in the same way as lvextend and lvreduce. See lvextend(8) and lvreduce(8) for more information."
[08:52] <jc_> tomreyn: okey. I will try that, and let you know. Thanks
[08:53] <tomreyn> jc_: if your goal is to allocate all the available space to this LV, then a better option than specifying the size using -L is to use -l 100%FREE
[08:54] <tomreyn> this ensures there's no leftover unallocated space (other than any that is strictly technically required)
[09:04] <jc_> tomreyn: thanks for your help. lvextend didn't help, as you suggested I have used lvresize and now I am able to see the space. Thanks for your time.
[09:04] <tomreyn> you're welcome
[09:06] <jamespage> sahid: I think I'm caught up
[09:23] <jamespage> sahid: no missed cinder - doing that now
[09:33] <sahid> jamespage: ack thanks
[09:33] <jamespage> sahid: no problem
[09:46] <jamespage> sahid: do I need todo anything with regards to DDT?
[09:47] <sahid> jamespage: i don't think so, it seems to me we are good
[09:48] <sahid> the package i would like you to double check is python-falcon, it's the one i'm currently working on and i did lot of changes
[09:48] <jamespage> sahid: https://launchpad.net/ubuntu/+source/python-ddt/1.2.1-0ubuntu1/+build/17174524
[09:50] <sahid> jamespage: https://launchpad.net/~sahid-ferdjaoui/+archive/ubuntu/eoan-train-proposed/+build/17177025
[09:50] <sahid> jamespage: let me double check
[09:50] <sahid> i might have not well updated the branch
[09:50] <jamespage> sahid: that's not been uploaded to eoan
[09:52] <sahid> jamespage: ok right, i remember now, there were an issue with the doc, i fixed it but corey asked me some extra work
[09:53] <sahid> i will ping you when it's all done
[10:13] <sahid> jamespage: cinder will not pass because of python-ddt, i just updated my repo, if you can review/sponsor it
[10:13] <sahid> https://code.launchpad.net/~sahid-ferdjaoui/ubuntu/+source/python-ddt/+git/python-ddt
[10:13] <sahid> https://launchpad.net/~sahid-ferdjaoui/+archive/ubuntu/eoan-train-proposed/+build/17188109
[10:27] <Greyztar> hello,i needed to run a script at startup,didnt find any rc.local but made one and it worked,is this ok to do or should use other means to run script?
[10:30] <tomreyn> Greyztar: a systemd unit file is a better approach.
[10:31] <supaman> Greyztar: unless you have changed something then systemd is managing startup, see here: https://askubuntu.com/questions/919054/how-do-i-run-a-single-command-at-startup-using-systemd
[10:31] <Greyztar> tomreyn: thanks also thanks supaman
[11:19] <jamespage> sahid: ok so I have a suggestion for python-ddt - use the orig.tar.xz from debian experimental
[11:19] <jamespage> rather than the upstream release tarball - that way you will have the docs I think
[11:21] <jamespage> will also make a merge/sync easier at some point in time
[11:21] <jamespage> dropping the -docs package is alot of divergence from debian
[12:50] <mason> Hey all. I have a VM here where qemu has this flag: "-global PIIX4_PM.disable_s3=1"  I don't see where that's controlled in the XML config, so I assume that's the default. Anyone know the proper knob offhand?
[12:53] <mason> Hm, and this might be a bug, as I'm told the knob is <suspend-to-mem enabled='yes'/> which I'm already doing.
[12:54] <sdeziel> mason: yes, that seems to be the right knob, works for me on Bionic
[12:56] <cpaelzer> also depends on the (virtual) chipset see src/qemu/qemu_command.c
[12:57] <mason> sdeziel: Bionic here. I'm not sure what the issue would be.
[12:57] <mason> ah, could be a chipset issue - I have very old hardware I'm using as hypervisors
[12:58] <mason> I'll migrate the thing to my desktop, which is skylake xeon
[12:58] <mason> (Not a live migrate and it doesn't copy configs, so it'll use the local native CPU type)
[12:59] <sdeziel> mason: I'd be surprised if this is related to the underlying host
[13:00] <sahid> jamespage: ack sounds good idea, there is a discussion in progress #ubuntu-devel
[13:03] <mason> sdeziel, cpaelzer: Oddly, I still see -global PIIX4_PM.disable_s3=1 with <suspend-to-mem enabled='yes'/>
[13:03] <mason> I'll try allowing both in the xml
[13:04] <sdeziel> mason: I'd double check (with virsh dumpxml) that the setting isn't drop after you're done editing.
[13:05] <mason> kk, good idea
[13:06] <mason> sdeziel: Spot on. How would that be rewritten? Is there a set of defaults that override configs somewhere?
[13:06] <TJ-> mason: did you edit the XML manually whilst the guest was active?
[13:06] <mason> TJ-: No. Guess was off.
[13:06] <sdeziel> mason: I know little about this area. I know there is an XML validation thing happening
[13:06] <mason> Guest*
[13:06] <TJ-> mason: because libvirt re-writes the XM on shutdown and I've been caught by that. using "virsh edit <domain>" is safest
[13:07] <mason> Ah, didn't know about that one. It validates on save?
[13:07] <sdeziel> yes
[13:07] <mason> nice
[13:07] <cpaelzer> on shutdown it transitions from active to inactive state, that is the write there
[13:07] <cpaelzer> dropping some details that are only known at runtime
[13:08] <mason> Does the config live somewhere other than /etc/libvirt/qemu? The edit shows me suspend to mem disabled, which isn't what was in /etc/libvirt/qemu. Surprised me.
[13:09] <cpaelzer> it usually is in memory
[13:09] <cpaelzer> the path is only a backup for restart and such
[13:09] <cpaelzer> not meant to be edited
[13:09] <cpaelzer> as sdeziel said, "virsh edit"
[13:09] <cpaelzer> TBH I never touched /etc/libvirt/qemu/*.xml - so I'd assume there never is a need to
[13:09] <mason> I was definitely editing with vi when the domain was down, but I'm happy with virsh edit if it validates.
[13:10] <sdeziel> mason: when I need to edit xml files by hand, I reload/restart libvirtd so that it picks up the updated files
[13:10] <mason> There we are. Got the correct flags this time. Now I'll try suspending the thing again. :)
[13:10] <mason> sdeziel: Oh, that's a good idea. Although I'll diligently use virsh edit now.
[13:11] <sdeziel> mason: yes, I only do hand edits when I need sed ;)
[13:11] <mason> aha, progress. It's yapping about not having the guest agent now.
[13:12] <mason> and, suspended - and now we'll see if it can wake up
[13:15] <mason> sdeziel, cpaelzer: If either of you has random free time, I'd be curious to know if you manage to have a guest sleep and then wake. We've gotten me to the same point where I was with Fedora's qemu-kvm, where I can tell domains to sleep, but it's the long sleep of death, never to awaken again.
[13:15] <mason> I'm trying to debug something vaguely unrelated, so I have no time pressure for this - it'd just be more convenient debugging on a VM than hardware - but it'd be cool to see this stuff work anyway.
[13:17] <cpaelzer> mason: I always used the xternal suspend/resume (virsh suspend ...)
[13:17] <mason> cpaelzer: virsh dompmsuspend --target mem here.
[13:18] <mason> cpaelzer: The RHEL 7 kernel has problems waking up my laptop when root is on md-raid. I can reliably reproduce it. It goes away with newer kernels, but it'd be nice to track the issue in the existing RHEL 7 kernel. If I can get it reproduced in a VM that'd be a nice convenience.
[14:27] <sahid> coreycb: anychance you upload: https://code.launchpad.net/~sahid-ferdjaoui/ubuntu/+source/python-mimeparse/+git/python-mimeparse?
[14:27] <sahid> https://pastebin.canonical.com/p/wQyntj7Sg2/
[14:44] <coreycb> sahid: couple of comments https://paste.ubuntu.com/p/3xh7rmfmXK/
[14:48] <sahid> coreycb: ack thanks
[15:33] <noregret> I have an ssh command that I would like to run on startup, what's the best way to achieve that? ubuntu server 18.04
[15:40] <blackflow> noregret: what command?
[15:52] <mason> noregret: Make a unit file for it. User services are probably the best way to do it, except, I haven't figured out how to get user services to automatically start.
[15:52] <mason> And a manual start defeats the porpoise.
[16:42] <noregret> blackflow: something like this: autossh -M41337 -f -o TCPKeepAlive=no -o ServerAliveInterval=15 -nNT -L 2234:localhost:8822 user@host
[16:43] <noregret> mason: unit file meaning a systemd service?
[16:53] <blackflow> noregret: yes. but I'd not run it as a user service, but as a regular system service, with a dedicated user for the tunnel
[16:53] <blackflow> newer systemd can even do dynamic users, so that's useful as well
[17:02] <TJ-> mason: you add the unit as a wants of (the user's) default.target
[17:13] <noregret> blackflow: thanks, i'll look it up
[17:16] <mason> TJ-: Ah, I'll try it. Thanks.
[17:17] <TJ-> mason: the all-user would be in/under /etc/systemd/user/{,default.target.wants/} or for per-user $HOME/.config/systemd/user/{,default.target.wants/}
[17:43] <mason> TJ-: I didn't know about the user default.target.wants, and didn't have it. Good to know.
[17:45] <TJ-> mason: default.target should be in /usr/lib/systemd/user/ and you create the .wants directories as needed in either /etc/systemd/user/ or $HOME/.config/systemd/user/
[17:46] <sarnold> is there a handy systemctl thingy that'll do that? or do you just need to do the symlink grubbing?
[17:47] <mason> sarnold: systemctl edit --user --force created the structure for me, so... Dunno. I'll experiment.
[17:48] <sarnold> mason: alright that's a promising start. I'm sure I'd forget the paths but stand a chance of discovering the edit --user --force again :)
[17:49] <mason> If I ever have to set systemd symlinks by hand that'll be "the line."
[17:50] <sarnold> the "why did we leave sysv?" line? :)
[17:52] <mason> The very same. :P
[18:01] <TJ-> sarnold: if the unit file has an [Install] section with WantedBy=
[18:02] <TJ-> see "man systemd.unit" -> "...Note: WantedBy= and RequiredBy= are used in the [Install] section to create symlinks in .wants/ and .requires/ directories...."
[18:03] <sarnold> someone should really redo all this with something simpler :)
[18:07] <mason> TJ-: Hrm, I had that. Half a sec and I'll find my userspace version.
[18:10] <mason> TJ-, sarnold: https://bpaste.net/show/300bfa1164c0
[18:10] <mason> TJ-: Or is the syntax for WantedBy different when it's a user service?
[18:11] <TJ-> mason: did you "systemctl daemon-reload" after adding the file ?
[18:11] <mason> TJ-: And where in all this is it a cleaner, more logical system? You don't have to answer this one.
[18:11] <mason> TJ-: No, but I was rebooting the VM, which presumably is good enough for that.
[18:13] <TJ-> mason: and did you enable the unit ("systemctl --user enable myunit.service") ?
[18:13] <mason> TJ-: Yes.
[18:13] <TJ-> mason: in which case the correct sym-link should have been created
[18:13] <mason> TJ-: And I could start/stop the service with systemctl. What wasn't happening was automatic start on boot.
[18:13] <mason> Half a sec and I'll reproduce the thing and show you the .config tree.
[18:13] <TJ-> mason: is user's default.target active ?
[18:14] <mason> TJ-: I don't know. Do I have to enable that?
[18:14] <mason> TJ-: The user lingers, but I wasn't aware of additional things that had to be done.
[18:15] <TJ-> mason: generally default.target should be active/enabled but I'm trying to check all the possibilities
[18:15] <mason> half a sec and I'll have a cleanly created .config tree to examine
[18:16] <TJ-> mason: are you creating this for a single user, or wanting it for all?
[18:16] <mason> TJ-: It's just one account starting the service.
[18:17] <TJ-> mason: if it's for all (/etc/systemd/user) then you need to "sudo systemctl --global enable myunit.service" ... if single-user, then "--user" for ($HOME/.config/systemd/user/)
[18:18] <mason> TJ-: https://bpaste.net/show/1cd5c64f56a0
[18:18] <mason> TJ-: So, is it unhappy with WantedBy=multi-user.target do you think? Is there something else I have to enable for user services to be started on boot, which I believe is largely their point?
[18:19] <TJ-> mason: as I said, and I thought you were using due to your pastebin, it HAS to be default.target - there is no multi-user.target for users (see "ls /usr/lib/systemd/user/ ")
[18:20] <mason> I'll try that.
[18:20] <mason> I'm assuming that implicitly starts after multi-user.target...?
[18:21] <mason> systemd service ordering is a mess
[18:22] <mason> Hm, default.target didn't do it either. Do I have to enable the user default.target explicitly or somesuch?
[18:24] <TJ-> mason: what does "systemctl --user show default.target" report?
[18:24] <mason> LoadState=loaded
[18:24] <mason> ActiveState=active
[18:25] <TJ-> mason: paste the entire report please
[18:25] <mason> TJ-: https://bpaste.net/show/f48288579686
[18:28] <mason> TJ-: Interesting. https://github.com/systemd/systemd/issues/2690 asks if the service is wanted by default.target, and mine isn't.
[18:28] <mason> TJ-: https://bpaste.net/show/08898c7469bf
[18:29] <mason> even after a daemon-reload
[18:31] <mason> If I'm reading that right, they closed that bug without resolving it.
[18:34] <blackflow> after some discussion and inability to reproduce it seems.
[18:34] <TJ-> mason: there's a linked commit that corrected the .config/systemd/user/ units which were set to Wants=multi-user.target, to be me default.target
[18:35] <TJ-> mason: what's your unit's name?
[18:41] <TJ-> mason: just to be clear; the user "seven" is logging in? Or, are you trying to get this job to run as the user "seven" as a system service even when that user has NOT logged int?
[18:49] <mason> TJ-: user is seven, unit name is sevendays.service
[18:49] <mason> TJ-: With the system service, yeah, it starts as that user even if that user isn't otherwise logged in.
[18:49] <mason> TJ-: As the system service, I assumed it would do the same - start on boot.
[18:50] <mason> sorry, s/system/user/
[18:51] <mason> TJ-: You're not saying that the service is supposed to only start when the user logs in, as a user service, are you?
[18:51] <TJ-> mason: ahhh, no, the point of --user units is to do something when that user logs in. pam_systemd starts systemd --user for the user
[18:51] <sarnold> ahhhhhhhhhh
[18:51] <mason> Well. That solves that then.
[18:52] <mason> TJ-: Thank you.
[18:52] <sarnold> I thought this was supposed to be similar to cron @reboot
[18:52] <mason> Likewise. But evidently not!
[18:53] <Odd_Bloke> Yeah, I found that out earlier this week when configuring user-level backup services.
[18:53] <Odd_Bloke> (For me, that's a definite feature, as there's no need to take new snapshots if I haven't even logged in yet; nothing should be changing without me. :p)
[18:53] <mason> Odd_Bloke: Did you drop back to doing it on the system with with a User= directive instead?
[18:53] <mason> Ah.
[18:54] <sarnold> mail delivery through procmail?
[18:54] <sarnold> crontabs?
[18:54] <sarnold> hmm, do both of those go through the same pam_systemd stack?
[18:55] <mason> Shush! Let him have his dream.
[18:55] <TJ-> mason: to run a service as a user its a regular system service unit with User=seven in [Service] - see "man systemd.exec"
[18:55] <mason> TJ-: Yeah, that's what I was doing before I decided to try it as a user service.
[18:56] <mason> c.f. the first pastebin
[18:56] <TJ-> mason: darn! round the world in 80 pains :)
[18:56] <mason> haha
[18:56] <mason> I thought I was doing it the wrong way. I have learned. Not wasted time!
[18:57] <sarnold> learned *and* taught :D double-win
[18:57] <mason> heh
[18:58] <TJ-> I've learned to... I seem to have become a systemd guru without knowing quite how!
[18:58] <TJ-> I wrote a fabulous systemd generator last week, to create .automount/.mount run-time units for each of several LVM LVs so they don't get mounted until accessed... works fantastically ... except at boot time, doh!
[18:59] <TJ-> since generators run before anything else, and the generator looks for /dev/$VG/* to get the LV names!
[18:59] <TJ-> at that point - LVM hasn't activated
[19:00] <mason> TJ-: Have the generator set up a watcher with inotify!!! \o/
[19:01] <mason> I'm not actually sure you can run inotify on a devfs, but, just saying.
[19:01] <mason> devtmpfs*
[19:30] <TJ-> mason: :) I just had to run the generator itself as a regular system service
[19:31] <TJ-> but, the exercise was fun - showed me the power of generators and how simple it is to create them
[19:36] <lordcirth> TJ-, that's interesting, I'll have to keep generators in mind
[19:37] <TJ-> lordcirth: for reference: https://iam.tj/projects/ubuntu/systemd-generator-LVM_automount
[19:37] <lordcirth> Thanks!