=== Dad is now known as Sebastien === lotuspsychje_ is now known as lotuspsychje === Sebastien is now known as Dad === Greyztar_ is now known as Greyztar [08:34] hi, can anyone help setting up gui in ubuntu server 18.04? i get to remote desktop screen where it shows me "xrdp - just connecting" popup with session dropdown having multiple options.. i tried alll of them but nothing seems to work :( [08:43] "systemctl status xrdp-sesman.service" shows this line "ExecStop=/usr/sbin/xrdp-sesman $SESMAN_OPTIONS --kill (code=exited, status=1/FAILURE)" [08:43] i am not sure if it is relevant [09:56] mehjari: a GUI is usually the main difference between a server and a desktop (there can be others, such as different hardware, workloads etc.) Do you really need a GUI on your server, or could you also manage it on a remote shell? [09:59] another option would be webmin, a web UI for managing systems. experienced system administrators would usually also frown at this, but might prefer it over a GUI. [09:59] tomreyn, i am trying to start a vagrant box in Ubuntu server, it is giving me SSH retry error. some online forums suggested that i should check what is the status of VirtualBox UI at that point, is there any popup etc. [09:59] therefore, as much as i hate installing GUI on server, i have to try GUI. [10:02] mehjari: I don't know about VirtualBox, but if you were using libvirt+KVM instead, then you could run virt-manager on a different box and connect it to the running libvirt instance on the server to see what the problem is - so no GUI needed on the server itself. [10:02] Does VirtualBox not have support for something like that? [10:05] rbasak: mehjari: I think you can expose the display [10:05] like you'd expose VNC on libvirt [10:05] and you can then attach to that [10:05] yeah, this looks good https://www.techrepublic.com/article/how-to-connect-to-a-virtualbox-vm-desktop-remotely/ [10:06] just the first hit on search engine, I did not try it myself [10:06] that would allow to see early boot of the guest and such [10:06] and IIRC virtualbox can control all on cmdline as well [10:06] although I don't know hot to enable the remote display feat there I'm sure that can be found === cpaelzer_ is now known as cpaelzer [10:09] if vbox is stuck at the popup "enable SLAT2" or VTIX technology in BIOS (which i want to rule out), the it doesn't show up with controlvm cli api as far as i can tell [10:09] i am not sure if they have logs which can reflect the state of UI at certain point [10:10] the Ubuntu VM is in Azure, and MSFT is charging monies, so i wanted a quick solution.. -.- [10:12] hello,is it normal that with snap apparmor isnt working,this is not on Ubuntu though but on Debian,just wonder in general if apparmor isnt ment to work on other installs with snap? [10:13] and also is it so that apparmor is sort of the sandboxing function of apps? [11:03] mehjari: in case this is still relevant, some folks in #vbox tend to be rather responsive. [11:05] Greyztar: /join #snappy [11:06] tomreyn: thanks i looked for #snap but didnt find any [11:50] tomreyn, thanks, will sync with vbox. :) === waveform_ is now known as waveform [14:53] Well, looks like Canonical might get a new customer....have a call scheduled with a SA today to disscuss Landscape === Wryhder is now known as Lucas_Gray [16:32] landscape is fun. :P [16:32] (I use it for all my new servers now xD) [16:50] teward, Isn't it expensive? Or is there a community version? [16:52] lordcirth: it does get expensive yes. (In this case I already have seats) [16:52] (and those're fixed-price not per-minute) [16:54] most of what landscape does is easily achivable with simple scripts [16:54] or packages already available in ubuntu [16:55] so just setup your ansible playbook to add that to your machines and - well - no problem [18:29] https://pastebin.ubuntu.com/p/dPQx4RyyzF/ [18:29] How can I figure out what's using all that memory? [18:29] Is that a kernel memory leak? [18:30] rbasak, highly unlikely. Are you low on memory? [18:30] https://pastebin.ubuntu.com/p/svVbxh7k9m/ [18:30] lordcirth: I am. My system has 32G, but it's thrashing. [18:30] https://pastebin.ubuntu.com/p/RD7sqVCQFR/ [18:30] No swap, but I added some to give me some headroom. [18:31] I see. Are you running any network filesystems, or anything which would keep a lot of files open? [18:31] No [18:31] rbasak, also, what Ubuntu version and kernel version? [18:32] 4.4.0-139-generic with livepatch [18:32] 16.04 [18:32] I could just reboot [18:32] But I don't like doing that [18:32] Because I'd like to fix the problem for next time too [18:32] This is a desktop btw. [18:32] rbasak, what's your uptime? [18:32] 175 days [18:32] That's pretty long for a desktop [18:32] What graphics drivers? [18:32] maybe the live patching messed something up? [18:33] AMD [18:33] My graphics setup is pretty non-standard, so a leak there could be likely. [18:33] 01:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Curacao PRO [Radeon R7 370 / R9 270/370 OEM] (rev 81) [18:33] Updating your graphics drivers without a reboot can cause all sorts of problems. [18:33] I'd just like to know how to pin the leak down to something, if there's any such mechanism possible. [18:34] That's beyond my skills. But I'd consider 175 days uptime on a desktop unreasonable and just reboot it [18:34] also, this isn't the proper channel to discuss desktop issues maybe ;) [18:34] sudo canonical-livepatch status --verbose # gives fixes "" ? [18:34] True, I didn't even notice [18:35] JanC: yeah, maybe :) [18:35] I don't need any desktop specific answer [18:35] maybe the kernel people can give some pointers [18:35] Just some help grokking kernel memory information, which isn't desktop specific [18:35] smem, etc [18:37] if status says livepatch: state: applied, you need to reboot anyway, and run ful-upgrade, there might be more kernels and patches waiting... [18:38] 175 days ... awesome [18:38] rbasak: wow; are those counts going up over time? [18:39] sarnold: AFAICT, yes. [18:39] rbasak, what is the output for inodes: df -i [18:40] OerHeks: looks OK. I don't want to paste that (privacy) but apart from snap loopback devices (squashfs), the highest IUse% is 36% for /, so I don't think it's that. [18:41] oke, just checking. [18:41] I appreciate the thought :) [18:42] then livepatch should be the culprit, AFAIK, waiting [18:42] Earlier I sent SIGSTOP to firefox and snap related process, so I can't run livepatch right now. [18:42] (since they were causing the thrashing) [18:43] I could resume the snap ones I guess now that I've added swap [18:43] I'm quite sure rebooting will fix the issue, but I'd prefer to find bugs before I remove all the evidence :) [18:43] it might be useful to talk to the kernel and graphics devs [18:45] Hmm. [18:45] "canonical-livepatch status" hangs, but I don't know if I caused that by interfering with it earlier. [18:46] rbasak: can you scan /proc/[1-9]*/pagemap to discover where the memory is allocated? [18:46] TJ-: I don't think it's allocated to any process [18:46] TJ-: based on smem output from https://pastebin.ubuntu.com/p/svVbxh7k9m/ earlier [18:47] rbasak: can you get bcc tools on this system? https://github.com/iovisor/bcc/blob/master/tools/memleak_example.txt [18:48] rbasak: if not, perf trace can probably do a similar job, but these are nice tools :) [18:48] sarnold: is that packaged? :) [18:50] https://github.com/iovisor/bcc/blob/master/tools/memleak.py [18:50] easy peasy [18:52] rbasak: ah, does "smem -w" confirm its kernel mem ? [18:53] Looks like it is packaged but only from Bionic :-/ [18:53] TJ-: right - it's under 'kernel dynamic memory' [18:55] rbasak: hey it is! :D bpfcc-tools [18:55] rbasak: haha, thanks, I hadn't realized it'd been packaged up yet. [18:57] rbasak: "smem -m | tail" might help a bit more [18:58] Biggest is 120 38624 4634930 [18:58] What are the units? [18:59] "smem -m | head -1" :) == " Map PIDs AVGPSS PSS " [18:59] I did look there! Those aren't units I understand. [18:59] The unshared memory (USS) plus a process's proportion of shared memory is [18:59] reported as the PSS (Proportional Set Size). The USS and PSS only include physical memory usage [18:59] "man smem" [19:00] Those still aren't units! [19:00] rbasak: not my fault! [19:00] OK, but my question still stands then :) [19:01] rbasak: I'd guess kB the same as free reports? [19:02] TJ-: as long as it's not "pages" or something, then that's only ~5G so not my leak :( [19:03] rbasak: it is infuriating when it doesn't say [19:03] Agreed! [19:04] rbasak: use "-k" it shows the units [19:05] 120 36.0M 4.2G [19:05] * TJ- hugs the man-page [19:05] rbasak: so no help there then [19:05] so gay [19:05] compdoc: if you are, so what? [19:09] I backported Bionic's bpfcc to Xenial. [19:09] But I get [19:09] $ sudo /usr/sbin/memleak-bpfcc -o 60000 [19:09] /virtual/main.c:18:1: error: could not open bpf map: Invalid argument [19:09] is maps/stacktrace map type enabled in your kernel? [19:09] It might be time to give up. [19:09] Thanks to everyone for trying :) [19:12] :-) [19:23] I just did a fresh install of lanscape on ubuntu 16.04, and when I try to register computers, they won't register and there's a "missing/invalid csrf token" error in the server logs.. google search turns up nothing.. any ideas? [19:25] maybe a strict setting in firefox/chrom(ium), https://get.todoist.help/hc/en-us/articles/208951085-CSRF-token-error-messages [19:26] OerHeks: the weird thing is that I get the error when trying to register a computer with the landscape-config command [19:26] so landscape client can't talk to landscape server .. getting CSRF error [19:27] mwheeler-: are you having to go through a proxy? [19:27] nope, no proxy, and both client and server are on the same L2 subnet [19:27] they're VMs on the same host [19:33] can you tell if the token is missing, or invalid - the latter could mean multiple connections causing confusion, the former I'm not sure what [19:34] good question.. I didn't pay that close attention to the tcpdump.. I'm blowing away the VM right now and rebuilding it just to see if there was something screwy with it.. it was my general "use and abuse" server when I needed to test something [19:35] mwheeler-: it sounds like the kind of error coming up from a library, most likely Python [19:37] TJ-: yeah, in the log message I see zopepublication.py .. so it is likely further down in the stack than landscape [19:42] mwheeler-: "zope" look to be the underlying framework they're using [19:43] yep [19:43] * TJ- recalls trying Zope some years ago and running away, scared [19:43] I feel that way about most web frameworks [20:02] CSRF token issues might be a result of an ad blocker or such (blocking cookies or whatever)