[00:24] <drab> I get the million <3
[00:24] <drab> it seems like if I disable predictable then the matching in udev is screwed up, don't have the time to debug why, but I can'tg et eth0 to become "lan"
[00:25] <drab> so if I make the names "predictable" again, and I will quote "predictable" to my death bed because it's the last thing those names are, the you can match with KERNEL="enp*"
[00:25] <drab> that seems to avoid the bridge device which wuoldn't match and therefore it all works in the end
[00:26] <drab> mega workaround but worth a million. at some point that systemd bug will make it through and it'll "just work" [tm]
[00:30] <OerHeks> drab, maybe this works, mask udev's rule file for the default policy: ln -s /dev/null /etc/udev/rules.d/80-net-setup-link.rules >> https://www.shysecurity.com/post/20160510-Ubuntu%20Network%20Interfaces
[00:39] <drab> OerHeks: that post itself says that method doesn't :)
[00:39] <drab> anyway it's fixed per above
[00:39] <drab> I'm gonna go enjoy the million dollar :P
[00:40] <drab> certainly hard earned
[00:42] <OerHeks> oh
[02:07] <DirtyCajun> Do we have a stable kernel release that works with AM4 ?
[02:07] <DirtyCajun> Reading the release notes it looked like 4.10 was promising but i wanted some opinions
[02:08] <sarnold> DirtyCajun: I suspect most folks just use xeons. You may have better luck in #ubuntu
[02:10] <DirtyCajun> That is the pleb channel. i stay far away
[02:11] <sarnold> hehe
[02:11] <DirtyCajun> Xd
[02:11] <patdk-lap> I dunno why you don't ask in the kernel channel
[02:11] <sarnold> I just figured it'd be more likely a gamer would know..
[02:11] <DirtyCajun> Wat. its not for gaming
[02:11] <DirtyCajun> its a 16 thread proc. its for my plex 2u
[02:11] <patdk-lap> heh? not for gaming?
[02:11] <sarnold> oh? I thought this was the new amd cpu; the only other person who's mentioned it on irc is a gamer..
[02:11] <patdk-lap> dunno what an AM4 even is
[02:12] <patdk-lap> people use amd in servers?
[02:12] <DirtyCajun> absolutely
[02:12] <patdk-lap> never seen one, ever
[02:12] <DirtyCajun> wel.... i work at a datacenter
[02:12] <DirtyCajun> and it is in about 40% of our blades
[02:13] <patdk-lap> I have racks of servers, not a single amd
[02:13] <sarnold> well what do you know, an amd chip in a server ;) http://www.thinkmate.com/system/stx-nl-qe12-2210
[02:13] <patdk-lap> :)
[02:13] <patdk-lap> they are always horrible for my performance workload, due to the lack of ondie l2 cache that intel has
[02:13] <DirtyCajun> patdk-lap, check out the specs of the Ryzen
[02:13] <DirtyCajun> and the upcoming server grade chips
[02:14] <patdk-lap> ya, I'll wait for real benchmarks
[02:14] <drab> why is mdadm and preseeding such a pain to work with :(
[02:14] <patdk-lap> as the benchmarks so far, are not accurate
[02:14] <DirtyCajun> passmark has 50+
[02:14] <sarnold> in fairness it's insanely hard to do benchmarking well
[02:15] <patdk-lap> the last benchmk I saw of amd new cpu line was, in this specific memory workload test, we beat the crap out of intel
[02:15] <patdk-lap> due to having double the amount of memory channels
[02:15] <patdk-lap> ya, anyone can make something beat the crap out of something else in one area, I want to know about normal workloads :)
[02:17] <sarnold> gah
[02:17] <sarnold> so, cpubenchmark.net numbers on it look pretty good but I was curious about other specs.. and was looking for an ARK link. sigh. :)
[02:17] <patdk-lap> I guess for a in memory database, that might be ok, if the ondie cache is acceptable to actually run the sql server
[02:18] <patdk-lap> hehe
[02:18] <patdk-lap> I'll wait a year or two atleast
[02:19] <drab> how si there not an option to nuke everything with preseed? just nuke, *everything*, don't try to mount raids and be smart
[02:19] <patdk-lap> intel has pretty much deadended their cpu lines though
[02:19] <drab> have I missed it?
[02:19] <drab> I keep getting crap because some of the disks had raid stuff on it that apparently mdadm --zero-superblock hadn't cleaned and install fails
[02:19] <drab> and it seems non-sense since I don't care what's on the disks, I just want the whole thing to be wiped and reinstalled
[02:26] <sarnold> patdk-lap: what I could find on the amd website is sorely thin on specs, but it does say for the eight core model, 2 channels of ddr4, 768kb of l1, 4mb l2, 16mb l3; 95W
[02:26] <patdk-lap> seems light
[02:28] <patdk-lap> man, have to go all the way down to a e5-2620v4 to find an 8core intel
[02:28] <patdk-lap> 20mb cache
[02:29] <patdk-lap> not as bad as it used to be
[02:29] <sarnold> I've got a e5-2630v3, very similar specs; but four memory channels, 40 PCIe lanes (instead ofthe 24 that other news sites were reporting for the ryzen)
[02:29] <patdk-lap> 4 memory channels
[02:30] <sarnold> so says ark, hehe :)
[02:30] <patdk-lap> so says my m otherboard :)
[02:30] <patdk-lap> so it *might* be comparable
[02:31] <patdk-lap> have to see what pricing turns out to be
[02:31] <sarnold> cpubenchmark said it'd be a bit cheaper than my cpu
[02:31] <sarnold> if you had compute-bound tasks rather than memory bound tasks, the ryzen might be a good fit
[02:32] <patdk-lap> ya, most workloads are memory issues
[02:32] <patdk-lap> except stuff like encoding video
[02:32] <patdk-lap> I don't know why that isn't memory bound, as I would think it should be :)
[02:33] <sarnold> it's nice to know amd's in the game again
[02:33] <patdk-lap> intel did take themselves out of it
[02:33] <patdk-lap> been years since they made progress
[02:33] <patdk-lap> and now they basically admitted to giving up
[02:33] <sarnold> they haven't had any real need
[02:33] <patdk-lap> sure they have
[02:34] <patdk-lap> they have almost let arm overrun them
[02:34] <sarnold> back in early opteron days
[02:34] <sarnold> but it's been a decade
[02:34] <patdk-lap> intel is getting backed into a server cpu corner
[02:34] <patdk-lap> with this, more and more amd systems for desktops, maybe laptops
[02:35] <patdk-lap> but more and more laptops are going arm by being replaced with tablets and phones and stuff
[03:41] <fishcooker> how to duplicate a server using the same configuration ?
[07:55] <ybaumy> hi. what can cause something like the following. on vlan 1 iscsi connections work like a charm. on vlan 500 iscsi connections work like a charm. but on vlan 501 iscsi connection are really slow and its a lottery if they work at all
[07:56] <ybaumy> there are no dropped packets. no denies on the firewall
[07:56] <ybaumy> i dont find anything at all
[07:56] <ybaumy> just that its incredible slow
[07:57] <ybaumy> all 3 vlans are using the same physical connection
[07:58] <ybaumy> the VM that runs vlan 501 has no memory or cpu problems
[07:59] <ybaumy> they all run 16.04 latest patches
[08:42] <ybaumy> i found the problem
[08:43] <ybaumy> fyi there was proxy arp enabled on the firewall for vlan 501
[08:44] <ybaumy> i found that in tcpdump there were no arp requests
[11:42] <a8o> Know any tricks for saving shell history as you enter commands?  I keep getting connections lost/dropped and i loose my shell history unless i exit normall.
[11:46] <hateball> a8o: first hit on google is http://askubuntu.com/questions/67283/is-it-possible-to-make-writing-to-bash-history-immediate
[11:47] <hateball> if you keep dropping your connection you may want to consider using mosh
[11:47] <a8o> cool thanks.
[11:47] <a8o> I use mosh on some systems, but not all can have it.
[11:57] <hateball> a8o: true enough
[12:01] <TafThorne> Try using screen or tmux on the remote end.  Then your shell session will persist even if you loose ssh connection to the remote.
[12:01] <a8o> never heard of tmux.
[12:02] <a8o> i do use screen alot, but some terminal apps get all screwy when I use them in screen.  But I do use screen for stuff I need to run and watch now and then.
[12:03] <TafThorne> tmux is very similar to screen.  Their use to be use case differences between then but now days I think both can do pretty much everything the other can do.  screen is often already installed on systems (just because it was around first I think).
[12:04] <TafThorne> As long as you do not resize the terminal I usually find things behave OK inside screen.  I do not use tmux so I cannot vouch for it.
[12:05] <a8o> if anything it's fun to try something new.  Lately I've been using the fish shell.  Love that one
[15:22] <jamespage> coreycb: hello
[15:22] <jamespage> coreycb: https://launchpadlibrarian.net/311293238/buildlog_ubuntu-zesty-amd64.vmware-nsx_10.0.1~git20170317.bfa48c0-0ubuntu1~ubuntu17.04.1~ppa201703171404_BUILDING.txt.gz
[15:22] <jamespage> does that seem familiar to you - I can't get vmware-nsx to build in a PPA
[15:22] <jamespage> coreycb: builds fine locally/on a cloud instance using sbuild
[15:24] <coreycb> jamespage, haven't seen that before.  i wonder if the tests do any db setup that is getting missed for some reason.
[15:30] <coreycb> jamespage, i see run_tests.sh has an option to recreate the database
[15:30] <jamespage> coreycb: yeah not sure about that
[15:31] <jamespage> coreycb: afaict nothing ever gets re-created on disk
[15:31] <jamespage> created rather
[15:31] <jamespage> coreycb: I lied as well - it failed on my laptop
[15:31] <coreycb> jamespage, oh..
[15:37] <drab> anybody here using ansible by any change and has it working with a preseeded install against the in-target chroot?
[15:59] <jamespage> coreycb: hmm
[15:59] <jamespage> coreycb: I might have it identified
[15:59] <jamespage> coreycb: why do we have a patch to make unit tests use an on-disk sqlite db?
[16:00] <coreycb> jamespage, is that in vmware-nsx?
[16:00] <jamespage> coreycb: no in neutron
[16:00] <jamespage> coreycb: disabling it to see if that helps
[16:01] <coreycb> jamespage, ok, not sure i'll take a look at it though
[16:36] <station> can ubuntu core be used on an intel pc with 4 Etherports as an gateway/rooter … firewall
[16:49] <drab> station: I'm doing that
[16:49] <drab> actually just getting through the bonding part on a second machine to get a "4gbit" link
[16:50] <drab> oh nm, you said core, assumed -server since this is -server, nm
[16:53] <station> drab: what did you use to set it up? is it a bad idea to alsow host like security cam directly on the rooter?
[16:54] <station> im new to server ....
[16:55] <drab> station: if by router you mean your internet modem, it's ok ime as long as you don't open it up and set up, say, port forwarding
[16:59] <station> drab: so i could set open sense / pfsense on one "roooter PC" and a separate server
[17:01] <drab> <troll level raised to yellow>
[17:01] <drab> station: what are you trying to achieve? I'm having troubles to understand what you're asking for
[17:02] <drab> are you trying to build a firewall for your home?
[17:05] <station> drab: im still learning about server the 4 port pc will replace a slow rooter and I have to get internet access to nextcloud  rocketchat …
[17:09] <station> was hoping to get them both on one device with ubuntu core but …
[17:20] <drab> nacc: you might know, is there any ML I can post to about say preseed questions? ubuntu-server seems to be development list only and not for support requests
[17:20] <drab> but I don't see any other place/ML where -server ppl are hanging out to ask a question
[17:25] <nacc> drab: you can ask here
[17:26] <nacc> drab: you can ask on -server too, you might just get redirected around a bit; i wonder if ubuntu-users is the right place
[17:30] <drab> nacc: I doubt -users deal a lot with preseed, feels like asking in ubuntu-desktop :P
[17:30] <drab> but maybe I'm wrong, not on -users
[17:30] <nacc> drab: what's your question?
[17:31] <drab> nacc: if you don't think -server ppl are going to hate me for the rest of my life for posting support requests to a -devel list then I'll try that
[17:31] <drab> I have two preseed problems
[17:31] <drab> both on -server and -desktop actually, altho latetr may be a bug in ubiquity installer
[17:31] <drab> so on -server, I can't figure out how to get preseed to ignore everything on the drive
[17:32] <drab> if I try to re-pxe a sytem that had raid on it, initrd I think will autostart the md device
[17:32] <drab> by the time parted kicks in the drives are in use and everything fails because there's no root device
[17:32] <nacc> pxe doesn't depend on a physical device being present for root
[17:33] <drab> I have to manually get a shell, mdmad --zero-superblock, stop the mdadm device, and then dd zeros for good measure
[17:33] <drab> sure, talking about preseed
[17:33] <drab> I can't basically reinstall a system automatically as it stands
[17:33] <drab> if the disks had stuff on it
[17:33] <nacc> while not having done this with raid, i've reinstalled many systems in the past that were already installed
[17:34] <nacc> drab: does it only happen with raid?
[17:34] <drab> even if they had a previous install, parted stops and asks me to confirm which device I wanna use and stuff
[17:34] <drab> nope, with a single disk too, same issue
[17:34] <nacc> drab: then you're not telling preseed enough
[17:34] <nacc> partman-lvm/device_remove_lvm boolean true
[17:34] <drab> I'm sure, I just can't figure out what I'm missing
[17:34] <nacc> partman-lvm/confirm boolean true
[17:34] <drab> lemme dpaste it
[17:35] <drab> http://dpaste.com/30B43FQ
[17:35] <drab> nacc: I have all of those already
[17:35] <drab> that's a grep partman srv1604.seed
[17:37] <nacc> drab: can you pastebin your whoel preseed?
[17:38] <nacc> drab: iirc, what you have is sufficient for everything to get wiped by the installer and if it's still prompting you for a selection, then something else is wrong
[17:40] <nacc> drab: and please be precise as to where it stops, becasuse "asks me to confirm wich device ... and stuff" doesn't help me undersatnd where the preseed is stopping
[17:40] <nacc> drab: ideally screenshots, i think
[17:44] <drab> nacc: ok, I'll get to that, gimme a few, thank you
[17:45] <drab> I should have a windows VM on vbox to repro this so that I can get actual sshots instead of blurry pics with my phone
[17:48] <nacc> drab: in the meawhile, can you pastebin your entire preseed?
[17:49] <drab> nacc: k, that I can certainly do right away, sec
[17:58] <drab> nacc: ok this has everything , line from pxelinux that I start the install with + srv1604.seed + the partman diff with the raid1 version
[17:58] <drab> http://dpaste.com/27KH6SP
[17:59] <drab> thank you for taking the time to look
[18:03] <nacc> drab: do you have multiple physical disks?
[18:07] <drab> nacc: in the server using the raid-1, yes, in the testbox using the simple one, no, just one ssd
[18:07] <nacc> drab: in the latter case, does it help to specify
[18:07] <nacc> d-i partman-auto/disk string /dev/sda
[18:08] <drab> ok, I'll add and try with that and see if it helps
[18:09] <drab> there is the equivalent of that in the raid1 already tho, and I still get errors due to raid being mounted, but that might be a diff issue
[18:09] <nacc> drab: right i think it is
[18:09] <nacc> drab: hence why i want to see where preseed stops :)
[18:09] <drab> ok
[18:10] <drab> I'm still working on getting a VM going, it's a bit of a multitasking morning with a few more interruptions than usual, appreciate your patience
[18:10] <nacc> drab: np, i might need to step away for a bit to run some errands/lunch
[18:11] <drab> no worries at all, thank you for your help
[18:12] <drab> btw I just figured out another neat trick this morning, just in case anybody cares...
[18:12] <drab> to do end-to-end ansible installs
[18:12] <drab> or anything else really, would be the same with chef, if you wanted to trigger a push and need ssh
[18:13] <drab> the late-command can install ssh in /target and then you can just start it and it'll work just fine so than you can ssh in/push stuff to it and make all the changes before rebooting
[18:14] <drab> plus put in the bios boot from hd first, second network, so now I can WOL machines at night, dd a few bits to the disk to remove the mbr, reboot, and it'll go into pxe, install and then ansible it
[18:14] <drab> poor man ipmi :)
[20:05] <rgb-one> Hey
[20:05] <rgb-one> any osTicket users here?
[20:09] <tarpman> oh wow. been years since I heard that name..
[20:09] <tarpman> rgb-one: not any more, and boy am I grateful for that!
[20:10] <rgb-one> tarpman, hehe why do ya say that?
[20:24] <drab> tarpman: is there a better alternative? I've been looking for a simple ticketing system for ppl to report problems, but didn't find much
[20:37] <rgb-one> drab, I found a list on reddit
[20:41] <tarpman> drab: I moved to request tracker
[20:42] <drab> rgb-one: link? I saw a whole bunch, but they are all way too complicated or kinda broken
[20:42] <tarpman> rgb-one: embarrassing... it's long enough ago that I'm actually having trouble remembering specifics
[20:44] <rgb-one> some free ones, Request Tracker and OTRS
[20:44] <rgb-one> https://www.reddit.com/r/sysadmin/comments/na2r0/to_my_fellow_helpdesk_operators_what_trouble/
[20:46] <rgb-one> Free ticketing systems: https://www.reddit.com/r/sysadmin/comments/2pp8op/free_ticketing_systems/
[20:51] <drab> rgb-one: if you go through that list it basically seems to come down to the usual 2 tho, OTRS and RT
[20:51] <drab> everything else isn't free (despite the topic, see Jira) or is the usual hosted web thingie
[20:51] <rgb-one> do you use either of those?
[20:51] <drab> I have in the past
[20:52] <drab> and they are too complicated for where I'm at now
[20:52] <rgb-one> Oh
[20:52] <drab> I almost just need a form with a listing at the back
[20:52] <drab> a mini django site would do I reckon
[20:52] <drab> even just sending me an email would work... I could do that to start actually
[20:53] <drab> I almost implemented it as a google form... I just need a few questions/guided step for ppl to report problem
[20:53] <drab> so that they don't write "my internet is broken"
[20:53] <drab> and it turns out they were at home... not joking
[20:54] <drab> I don't need multi user, report generating ITIL TLA MSG BLAH BLAH super duper flashy mobile first
[20:54] <drab> :P
[20:54] <drab> but it seems I'm a minority so there's no such thing
[21:05] <rgb-one> I see
[21:11] <drab> nacc: this was my other problem btw, a bug still open since 2011... https://bugs.launchpad.net/ubuntu/+source/netcfg/+bug/713385
[21:11] <drab> still working on those sshots
[21:12] <drab> I have the same prob as the initial guy, server with a quadport nic
[21:13] <nacc> drab: and they don't report link up even though they have link?
[21:17] <drab> the problem is with the "predictable" names, in some cases the "first" port is not the one I have pxe set up on
[21:18] <drab> I understand the theory about predictable naming, but honestly I never had an issue with eth* stuff coming out of order and I could reliably say to pick, say eth4 for pxe
[21:18] <nacc> drab: and do you pass BOOTIF= down?
[21:19] <drab> so I used ifnames=0 to go back to that, which worked and could reliably pxeboot stuff, but that broke everything when I started working with the new VM server and need to preconfigure bridges
[21:19] <drab> have not used BOOTIF=, but I passed netcfg/choose_interface=eth0 if it means the same thing
[21:20] <nacc> well, if you use BOOTIF= netcfg/choose_interface=auto should work
[21:20] <nacc> fi you don't use BOOTIF=, then yes, you might need to specify the network interface
[22:14] <nacc> drab: if you don't use your recipes (but use 'atomic' or something) does it work?
[22:15] <drab> nacc: I did some more tests... it works with auto, but it doesn't do what I'd like it to do. basically, for whatever reason, what by the rear panel are port 3 and 4 come up "earlier" with systemd and are named enp129XXXX Vs the other two enp4XXX
[22:16] <drab> so I guess they are selected first and stuff gets installed/set up with the wrong interface
[22:16] <nacc> strange
[22:17] <drab> the "funny" thing is, if I disable predict interafces with ifnames=0
[22:17] <drab> then the enp4XXX gets eth0 etc
[22:17] <drab> so eth0-3 match the panel layout
[22:17] <drab> and I get preseeding etc on eth0 which is what I want
[22:18] <drab> but like I said if I do that then I get other problems when I get to setting up the bridge
[22:19] <drab> at this point I'm inclined to give up and work around it with ansible
[22:20] <drab> as long as the install works with predictable names, when the box comes back I can reconfigure it
[22:21] <drab> the thing that makes even less sense is that the dhcp server is set to tftp *only* on that first interface, the others don't
[22:22] <drab> so it pxeboot from iface1, then when installer stuff it loaded order is rearranged by systemd or whatever
[22:22] <drab> it dhcp out on iface3, gets an ip and continues the installation from there
[22:23] <drab> I guess a good fix would be to have the installer to use the same iface it pxebooted on, but it might not know what that is
[22:24] <drab> nacc: I'll test with atomic
[22:29] <drab> nacc: nope, same error, no root file system defined
[22:31] <leetskeet> hello all, recently setup server 16.04 and I have an issue where after a reboot, I have to complete a local login before SSH with key login will work
[22:32] <leetskeet> ran systemctl enable ssh and still same issue
[22:39] <leetskeet> anybody around
[22:58] <drab> ok so I found the error...
[22:59] <drab> nacc: ubiquity: error ddf1 wrong # of devices in RAID set "ddf1_kvm5" on /dev/sda
[22:59] <drab> I've no idea why I get that
[22:59] <drab> there's no raid on that disk, it's an old windows disk, or maybe windows was installed with fakeraid and that did something to it...
[23:08] <JMichaelX> did a release upgrade a few weeks ago on home server, from 14.04 to 16.04, and just noticed that installation is using kernel 3.13.0-110-generic
[23:09] <JMichaelX> desktop on 16.04 is using kernel 4.4.0-66-generic. is there a reason for this difference?
[23:10] <genii> You haven't done a dist-upgrade
[23:10] <drab> JMichaelX: 16.04 server uses 4.4.65 too, prolly just missing a dist-upgrade
[23:13] <sarnold> JMichaelX: how did you do the upgrade?