[04:37] rbasak: specifically waas pinging that in my latest branch, which tries to speed up the importer (and solve our idempotency issue), we would do branch manipulations at the end, so the tests need to change a bit (import_uunapplied_dsc, e.g., no longer does any branch manipulation). Will take some more time for me to rework the tests (I have all the tag parts of the tests passing now. [05:15] Hi [05:15] Who know how to solve this problem, why forever doesn't run from crontab but works from terminal? [05:15] https://github.com/foreverjs/forever/issues/997 [05:15] this is my theme [05:46] PATH [05:46] we can't run command in crontab without new env variables [05:47] yes, I've solved it. are you interested what I did? [05:47] add new evarinment variables [05:48] and explicetly say cron to use bash [05:48] said [05:54] I've solved this, there is decision https://github.com/foreverjs/forever/issues/997 [09:02] I realize I'm too busy when I start music after working for 4 hours already :-/ [09:02] good ?morning? [09:05] good almost noon. and lol yeah, been there :) [11:21] my google-fu is failing me, is there a ubuntu-specific debsecan alternative thingy? [11:25] What's debsecan? [11:25] Oh. http://www.enyo.de/fw/software/debsecan/ [11:25] I'd never heard of that [11:26] it'd be cool to have a tool like that for ubuntu. supposedly debsecan should be easily modifiable to read ubuntu specific databases [11:26] I don't know of any equivalent. You might want to ask in #ubuntu-hardened where the security team hang out [11:26] ohhuh, yeah, I thought this was -hardened :) thanks. [12:11] i am trying to get tgt to show a lun to my esxi host that is a LVM lv. When i give tgt a .img (or whatever) that i create with dd it works just fine. But any time i try to give it a disk (/dev/sdXX) or a lv (/dev/mapper/xxx/xxx) it doesnt show on the esxi side. any thoughts? [12:45] i asked the question for 2 weeks. Never got an answer... and my server magically decided to like it and start working... mind blown. [12:47] DirtyCajun: zen of sysadmin. any problem will eventually fix itself. :) === miguel is now known as Guest89445 === tyhicks is now known as Guest86792 [14:18] hi [14:18] Who can estimate my English speaking? Can I make video tutorials. I'm going to do tutorial about linux? [14:19] this is my test video https://www.youtube.com/watch?v=fPAUiqEsBuU&t=2s [14:27] neo4: wrong channel [15:03] hey guys, anybody got a if-pre-up.d script to work on 18.04? [15:06] Hi [15:06] hi [15:07] Is it the right place to ask about systemd? [15:07] TvL2386: see https://wiki.ubuntu.com/BionicBeaver/ReleaseNotes#Network_configuration [15:07] TvL2386: AFAIK, since networkd doesn't support blocking pre-change hooks, as long as you use systemd-networkd you can't have those. [15:08] Tomer: you can ask about systemd _in Ubuntu_ here. [15:08] I'm trying to understand how a service is being run as a user [15:09] the systemd directive is User=blah [15:09] I did not see it used anywhere [15:09] rbasak: yeah I put my executable bash script in /usr/lib/networkd-dispatcher/routable.d and expected it to be ran on boot, but it's not [15:10] Tomer: are you looking for systemd user instances? [15:10] postgres, for example, has a user in the shadow file, but no User directive anywhere. The process is being run as a user [15:10] TvL2386: did you install networkd-dispatcher? [15:11] rbasak: networkd-dispatcher 1.7-0ubuntu3 was already installed [15:11] Tomer: it's quite common for daemons to drop their own privileges down from root. I don't know about postgres specifically without looking, but in these cases there would be no User directive in systemd. It'd be down to individual daemon configuration. [15:11] I have a correct #!/bin/sh shebang, line 2 is `date > /tmp/iptables-restore` but that file is never created... so it's not ran? Also cannot find any evidence in /var/log/syslog that it was run [15:12] I've searched the source code of postgres, there's no use of setuid, setguid or seteuid. [15:12] Each deamon is supposed to drop it's own privileges? There's no implicit mechanism for that? [15:13] systemd implicitly create a cgroup per daemon, btw [15:14] Where can I ask where the postgres permission are dropped? [15:14] just a guess: #postgres [15:14] Traditionally all daemons dropped their own privs [15:15] Nowadays systemd can do it for you, but it's more complex in terms of the privileged operations daemons need to do first, so not all of them do it that way yet [15:15] (and may have no plans to change) [15:16] just checked nginx for fun. The master process is ran as root, the workers www-data [15:17] same thing for php-fpm [15:20] Ok, thanks! I'll ask the postgres guys as well [15:24] Hi there, I'm trying to wrap my head around netplan. How would I get the equivalent of this? < https://pastebin.com/8ff6953V -- e.g. both dhcp and a static IP on an alias of the same interface. The examples I can find show 1 or the other, but not both? [15:26] good luck :) [15:26] I've copied my executable shell script to /usr/lib/networkd-dispatcher/*.d/ but it does not get executed on boot [15:27] it should work according to https://wiki.ubuntu.com/BionicBeaver/ReleaseNotes#Network_configuration [15:27] so how do you purge `netplan`? [15:34] cyphermox: see hackeron's request above - I can't find any obvious answer based on the docs for that use case. [15:35] I wonder if networkd would stomp on extra IPs if DHCP is enabled regardless? I don't know. [15:35] TvL2386: netplan does nothing if no netplan config is defined, AIUI. So just install ifupdown if you don't want to use netplan. [15:36] hackeron: the obvious syntactic answer would be to enable dhcp and also define addresses. I don't know whether that will work though. [15:36] hackeron: in fact [15:36] hackeron: "Add static addresses to the interface in addition to the ones received through DHCP or RA." [15:36] hackeron: so there's your answer [15:37] From https://netplan.io/reference#common-properties-for-all-device-types [15:37] rbasak: no netplan config? My /etc/netplan/01-netcfg.yaml contains the correct configuration to bring up ens160 and works fine [15:38] so how do you purge `netplan`? [15:38] that was just a joke :) [15:38] rbasak: ah, that did work! :D [15:38] growing pains here [15:39] rbasak: easy enough - I'm trying to embrace netplan, it's a bit frustrating though - for example installing netscript breaks openvpn and all kinds of other things, so removed that and trying to use netplan exclusively :P [15:39] I'm either screwing up or there's a bug with that networkd-dispatcher not being executed or not executing my scripts [15:40] hackeron: feedback on breakages appreciated. Though some of it may be networkd limitations rather than netplan - for example on the hooks thing - so that might be something that networkd upstream needs to tackle rather than netplan. [15:41] TvL2386: I don't see anything obvious that you're doing wrong there. [15:41] TvL2386: networkd-dispatcher acts on dbus signals so perhaps use a dbus debugger to see if the signal is being generated at all? [15:42] (or hack networkd-dispatcher to give you that debug info) [15:42] rbasak: thanks for the help anyway ^^ [15:43] rbasak: I thought DBUS was a ubuntu desktop thingy -_- [15:44] It's not Ubuntu-specific at all. Is it a freedesktop thing? I'm not sure. [15:44] I do quite like it as a language agnostic but powerful enough IPC mechanism. [15:44] Powerful = signals, properties, etc. [15:49] anyway: is it a desktop thing or also a server thing? [15:49] dbus-daemon is running [15:50] on my server [15:53] It's much more common on the desktop, but it isn't exclusively desktop. [15:53] systemd uses it quite a bit === tyhicks is now known as Guest87080 [17:47] rbasak: one difference in buildflags between debian and ubuntu is -Wl,-Bsymbolic-functions (ubuntu has it, debian doesn't) [17:47] does that raise a flag? [17:47] I've yet to check what it means [17:48] oh, it does mention "global symbols" [17:48] and the segfault is exactly about a global variable that suddenly became NULL [17:51] ...and I see this in d/rules [17:51] LDFLAGS="$(dpkg-buildflags --get LDFLAGS | sed -e 's/-Wl,-Bsymbolic-functions//')" \ [17:51] hmmm [17:51] and [17:51] - Add LDFLAGS to avoid issues with accessing global variables in [17:51] shared libraries (LP #1470687, LP #1472115). [17:51] Launchpad bug 1470687 in autofs (Ubuntu Vivid) "Not working properly with compile options "-symbolic-functions"" [Undecided,Fix released] https://launchpad.net/bugs/1470687 [17:51] Launchpad bug 1472115 in autofs (Ubuntu) "autofs start fails: segfault at 0 ip 00007f738cb881bb sp 00007ffeff888f70 error 4 in lookup_file.so[7f738cb76000+2b000]" [Undecided,Fix released] https://launchpad.net/bugs/1472115 [17:52] heh [17:52] definitely looks familiar [17:54] ah, I see it [17:56] fixed [17:56] nice [17:56] * ahasenack commemorates with some coffee [21:45] hi [21:46] the official Vagrant box at https://app.vagrantup.com/ubuntu/boxes/xenial64 enables the serial console [21:46] by having this in its Vagrantfile [21:46] vb.customize [ "modifyvm", :id, "--uartmode1", "file", File.join(Dir.pwd, "ubuntu-xenial-16.04-cloudimg-console.log") ] [21:47] what's the reason for that? [21:47] headless console output? [21:48] I just tested "--uartmode1 disconnected" and it boots 3x faster [21:51] anyone here using seafile? [21:51] https://paste.kde.org/pz4qubrek (running 5 times each) [22:16] blackflow: might be worth asking them about it -- they may enable serial consoles as a matter of course, just because the time when you absolutely need one is a bit too late to try to set one up.. === hehehe is now known as Guest15396 === Guest15396 is now known as hehehe [22:53] sarnold: I suppose that's for nicolas17? he asked about why it's on by default. and I concur. [22:53] the slowdown is pretty terrible, I wonder if it's a VirtualBox flaw [22:54] some of the log file looks corrupted too, like it drops characters sometimes [22:54] sorry blackflow, d'oh [22:54] but serial console output is indeed slow. FreeBSD recently made some patches which radically improved bootup speed on AWS, merely by not outputting anything by default (like grub's "quiet") [22:54] ugh AWS [22:54] they recently added real-time view of the serial output [22:55] previously it only captured the screen after state changes === hehehe is now known as Guest61509 [22:56] I don't think it's VBox fault. That thing is a blast from the past, it actually has baudrate, default I believe is 115kbps [22:56] tail -f on the log file seemed to go at tens of characters per second [22:57] and corrupting some text sometimes [22:57] audit: nitializing netlink subsysdibl) 6.87] di te=00ud(18308791)iniaze === Guest61509 is now known as hehehe