/srv/irclogs.ubuntu.com/2016/04/08/#ubuntu-server.txt

hallynrbasak: great, thanks00:43
patdk-lapheh, I always set dedicated ipmi interface, set static ipmi ip01:52
patdk-lapnormally fixes up those issues01:52
sarnoldpatdk-lap: even for the home lab?02:19
patdk-lapexpecially02:26
patdk-lapI'm running 7 vlans currently here at home02:26
patdk-lapnone of that is a lab though02:27
sarnoldnone of my switches can do vlan, heh02:38
sarnoldI've never learned enough about vlans to see it as a priority, so i've never spent the extra money02:39
patdk-lapI don't own a switch that doesn't do vlans02:39
patdk-lapswapping the current one out soon, for a poe model02:40
patdk-lapcrap, this unbound issue I have, is still present as a bug :(02:42
=== prince is now known as TheRaiser
=== TheRaiser is now known as prince
=== prince is now known as cholo
=== cholo is now known as meu
=== meu is now known as ShortCircuit
=== ShortCircuit is now known as coolwhip
=== coolwhip is now known as ``
=== athairus is now known as afkthairus
=== strigazi_AFK is now known as strigazi
fricklerjamespage: coreycb: https://bugs.launchpad.net/ubuntu/+source/nova/+bug/156781107:48
ubottuLaunchpad bug 1567811 in nova (Ubuntu) "nova-compute should depend on libvirt-bin.service instead of libvirtd.service " [Undecided,New]07:48
jamespagefrickler, oh good spot!07:48
smbjamespage, if possible maybe depend loosely on both variants. Would be nice to have something like "provides" aliases for services. But I cannot find anything07:57
jamespagesmb, we only have libvirt-bin in Ubuntu right?07:58
smbjamespage, right now, but it might change in future when we get rid of more delta07:58
jamespagesmb, sure - we can switch when we do that :-)07:59
smbjamespage, of course :) but still a pain to always think of the implications if say backporting into cloud-archive08:00
smbfor sysvinit scripts it was at least possible to say this provides libvirt-bin and libvirtd... but rather useless nowadays08:00
=== kickinz1|eod is now known as kickinz1
fricklerjamespage: smb: if you want to add Alias=libvirtd.service to libvirt-bin.service, that would also work, just tested that variant08:41
jamespagefrickler, that's a nice backwards compat feature08:42
smbYeah... wondering why I did not see that in the systemd.service doc08:43
smbjamespage, would you know whether hallyn has some other things queued or shall I wait till he comes around to sync up08:44
jamespagesmb,  not sure tbh08:44
jamespagesmb, I'd wait for hallyn08:44
smbjamespage, ack08:44
jamespagesmb, I'd also say that this is not critical for xenial08:44
jamespagehmm - well not from openstack's perspective08:44
jamespageI have no idea about other systemd units that might be dependening on libvirtd08:45
smbjamespage, hard to say. just if there is an "Alias" option that should not break anything existing while resolving anything we do not know about08:45
jamespagesmb, yah so maybe it does make sense for compat with direct syncs from debian08:46
jamespagesmb, shall I raise a libvirt task for that bug as well?08:46
smbjamespage, That is what we usually tried to maintain (for build dependencies and in sysvinit, just did not know about alias for systemd). yeah I think makes sense. probably can be lesser prio as you fix nova08:47
bvianyone experience with ServeRAID M5120 in jbod mode?08:52
bviinstaller doesn't even boot when this hba in in the server (lenovo x3650m5)08:52
smbfrickler, Just to confirm: under which section did you add the alias? Install?08:55
fricklersmb: yes, then disable/enable the service once in order to create the symlink08:57
smbfrickler, ok, thanks08:58
bvierr M5210 .... damn typo09:09
fricklerwould it make sense to change to upstream connection of https://launchpad.net/ubuntu/xenial/+source/nova to mitaka instead of essex?09:09
fricklerjamespage: https://bugs.launchpad.net/ubuntu/+source/keystone/+bug/156788110:53
ubottuLaunchpad bug 1567881 in keystone (Ubuntu) "keystone postinst should not call keystone-manage pki_setup" [Undecided,New]10:53
=== `` is now known as prince
BlessJahubuntu seems to completely ignore preseed file I provide12:30
BlessJahIn syslog I can see file has been loaded properly, but I cannot get any directive working, it still asks me questions that has been answered12:32
fricklerjamespage: https://bugs.launchpad.net/ubuntu/+source/nova/+bug/1567935 is my final one for today, but https://bugs.launchpad.net/neutron/+bug/1567923 might also be interesting for you13:13
ubottuLaunchpad bug 1567935 in nova (Ubuntu) "nova-compute-libvirt should not depend on open-iscsi" [Undecided,New]13:13
ubottuLaunchpad bug 1567923 in neutron "Neutron advertises too high MTU for vxlan" [Undecided,In progress]13:13
supNowjust upgraded ubuntu server to 14.04 and I'm getting a grub_term_highlight_color not found error. google search suggests it can't find grub but I havn't found any real information on how to fix.13:27
supNowFrom reading it's saying that it can't find the bootloader. I have booted into a live session from a usb drive. Is there an easy way to fix it from there? Or should I try to boot into ubuntu-server using grub from the usb drive?13:28
hateball!fixgrub | supNow13:30
ubottusupNow: GRUB2 is the default Ubuntu boot manager. Lost GRUB after installing Windows? See https://help.ubuntu.com/community/RestoreGrub - For more information and troubleshooting for GRUB2 please refer to https://help.ubuntu.com/community/Grub213:30
supNow@hateball it was actually ubuntu-server upgrade. There is no windows. Will that same thing work though that you linked?13:32
hateballsupNow: Yes, it just tells you how to reinstall grub13:35
supNowok great, thanks13:35
coreycbjamespage, we're not going to be able to sync magnum at this point in the cycle as it needs a big bump to python-docker13:47
hallynsmb: queued up for libvirt? i got nothing13:47
ddellavcoreycb swift is ready for review/push lp:~ddellav/ubuntu/+source/swift13:47
coreycbddellav, ok13:48
jamespagecoreycb, 1.5.0 not good enough?13:48
jamespagecoreycb, we may want the new version anyway - jgrimm - thoughts ^^ ?13:48
coreycbjamespage, not according to requirements.txt13:48
smbhallyn, k, I sent you a patch as well as have pushed a branch to lpgit. don't think its urgent now that nova got fixed13:48
jgrimmjamespage,  syncing magnum?13:49
jgrimmor bumping up python-docker?13:49
coreycbjgrimm, python-docker13:49
jamespagejgrimm, no syncing python-docker - we're right up to date with docker so we may actually want that right?13:49
jgrimmyes, we are at latest upstream13:50
jgrimmnot sure if python-docker is that current tho?13:50
jgrimmdocker.io-1.10.3 is what we carry.13:51
coreycbjgrimm, it's a bit behind, we're currently at 1.5.0 and it looks like they're up to 1.8.0 upstream13:51
hallynsmb: you didn't push to the xenial branch?13:52
smbhallyn, no, though I could have I guess. I used a smb/xenial one13:53
jgrimmcoreycb, indeed 1.5.0 sounds quite long in the tooth.   I'm not seeing anything in their docs that gives me a warm fuzzing on what versions of docker their client supports?13:57
jgrimmcoreycb, but i'd think you'd be better off moving up, though testing needed13:58
kaffieni ran a do-release-upgrade and ended up on 16.04.  Now our veeam setup cannot contact the NFS shares.  Is there something i need to configure on my ubuntu server to allow that again?14:10
kaffienveeam cannot connect to the SSH server i have running either14:12
kaffienbut i can via a ssh client  on that same workstation.   nmap shows all the appropriate ports are open.14:12
kaffiennm wrong channel14:12
coreycbjamespage, ddellav: all of the final mitaka release packages are uploaded to xenial for mitaka now (sans the magnum sync)14:28
jgrimm\o/14:31
=== afkthairus is now known as athairus
ppetrakiif I use an lxd profile to limit the number of cpus to say 2. /proc/cpuinfo is showing two cpus, but lscpu is showing all the host cpus. Is that the intended behavior? http://pastebin.ubuntu.com/15690427/15:38
patdk-wkyes15:39
patdk-wkthe kernel restricts what cpu's things can run on15:39
patdk-wkbut it doesn't go around limiting the view of the physical hardware15:40
ppetrakiah ok15:40
ppetrakithanks15:41
sdezielppetraki: /proc/cpuinfo of the container is a lxcfs mount showing just the CPUs that you have enabled for that host15:45
naccppetraki: lscpu is looking in /sys/ in contrast, iirc15:50
tiblockHi. I have ubuntu 14.04 server with 20GB on / and some logs files ate everything and i was not able to lauch anything so i removed about 18gb logs but `df -h` says only 300mb free, `ncdu` says 2.1gb used. Tryed to do `tune2fs -m 5 /dev/...` but its not helping. How to get space back?15:58
tiblockOh, it just jumped to 9.8gb free... Okay. Problem kinda solved itself. Still missing about half but 9.8gb its okay.15:59
nacctiblock: might need to issue a `sync;sync` if you deleted a ton of files15:59
nacctiblock: it would probably be better to figure out what log exploded15:59
lordievadertiblock: Restart your loging service, it probably still has a link open to those files, thus they are still on disk.16:00
tiblocknacc, after `sync;sync` still says 8.8gb used and i fouln log, my own application made it, forgot to add folder to logrotate. I deleted that logs.16:00
nacctiblock: ah ok, yeah i'd check lsof or as lordievader said16:01
nacctiblock: in case it's being held open16:01
nacc`du -h` at the appropriate level might give you a better idea of what is consuming space rather than `df`16:01
tiblockAh i see, `lsof` shows 7115024813 byte log file opened and says `(deleted)`. Okay, there is my space. Thank you very much!16:02
tiblockAwesome! After restart of that application i got all my space back.16:03
tiblock2.2gb used16:03
nacctiblock: nice, yeah, it was presumably still ref'd so couldn't be fully removed from disk (just unlinked from the fs effectively)16:05
lordievaderYou can do lovely tricks with this...16:06
nacclordievader: "lovely" :)16:08
sdezielnacc: https://bugs.launchpad.net/ubuntu/+source/unbound/+bug/1556308 was approved for upload when you get a chance16:13
ubottuLaunchpad bug 1556308 in unbound (Ubuntu) "[FFe] Please merge unbound 1.58-1 from Debian unstable" [Wishlist,Triaged]16:13
naccsdeziel: ack, i don't have upload rights :)16:14
naccsdeziel: hence sponsors is subscribed16:14
sdezieloh didn't know. Thanks anyways16:14
Bluekingpatdk-wk  everything works fine now :)16:14
patdk-wkheh, I just filed an unbound bug last night, that just annoys me16:16
sdezielpatdk-wk: yeah, noticed this16:16
patdk-wkstupid unbound segfaulting on libcrypto :(16:17
naccsdeziel: np, i'll see if i can find somebody16:17
sdezielpatdk-wk: would you mind opening a bug in Debian re /usr/sbin being in PATH? The Ubuntu delta should be kept minimal16:17
patdk-wkfigured cron restart would be a temp fix, nope :(16:17
patdk-wkah, I'll look16:18
sdezielpatdk-wk: latest Debian and soon Ubuntu will use the package helper which now calls unbound-anchor with a full path. Only unbound-checkconf will need fixing making it a oneliner16:18
=== strigazi is now known as strigazi_AFK
ppetrakibefore first install lxd, I configured my own bridge, bridge0 (via nm) and made that the default bridge. Then I created a new profile called scaleout, which when used makes my containers use lxcbr0, which doesn't even exist. Even claims there's an IP address, which is unreachable. That's a bug right?18:16
sarnoldppetraki: that does sound like a bug; iirc there's a way to tell lxd which bridge to use (it's liable to be a different mechanism than whatever you used to mark yours 'default') -- but it feels funny that it then reports unreachable IPs18:44
ppetrakisarnold, well, I told it not to even make the lxcbr0 (whatever the default bridge is) to begin with, it physically doesn't exist. So it's like the new profile, which I created from scratch decided to inherit a device config without telling me. It gets better, when I take the 'device' config from default and insert it into my new profile, those containers now have no interface.19:03
ppetrakisarnold, about to retest on a fresh vm, I'll keep you posted19:04
sarnoldppetraki: note that recent lxd changes have switched to using lxdbr0 by default, among other bridge-related changes https://launchpad.net/ubuntu/+source/lxd/+changelog19:05
sarnoldppetraki: it may be worth mkaing sure your image has the 2.0.0~rc9-0ubuntu3 release..19:06
ppetrakisarnold, ok, I have that release installed. the default config isn't getting my bridge. http://pastebin.ubuntu.com/15696446/  . just installed this vm from scratch19:18
sarnoldppetraki: probably one for stgraber ^^^19:20
ppetrakisarnold, ... and if I restart the lxd service it's there :)19:20
* ppetraki sees if it can use it19:20
sarnoldhunh19:20
ppetrakisarnold, http://pastebin.ubuntu.com/15696528/19:21
stgraberppetraki: either you're on an old packaging which doesn't restart lxd-bridge for you or your lxc profile command ran right before the change is effective19:22
sarnoldppetraki: try the ps aux again after sudo service lxd start; iirc it's a socket-activitated thing19:22
ppetrakistgraber, just installed the whole distro an hour ago19:22
stgraberppetraki: ok, so chances are that if you ran that lxc profile command again a second later it would have been right19:22
ppetrakistgraber, see http://pastebin.ubuntu.com/15696446/19:22
stgraberppetraki: ok, so yeah, you are on the latest, which means that the change very likely was applied a few ms after you ran that lxc profile command19:23
stgraberppetraki: profile changes can only be applied after the daemon is started so the very first query you do may return the old config. We do however hold container startup until the config was updated so you can't race it and start a container with the old bridge.19:24
ppetrakistgraber, I'm not the fastest typist in the world, perhaps a few seconds passed between the start and me asking for the profile. In either case, it's easy enough to sleep a sec or restart the service before continuing.19:26
stgraberppetraki: lxd is socket activated19:26
stgraberppetraki: so the first lxc command you run will start it, it won't be running until then19:27
ppetrakistgraber, ah, right, the new systemd stuff19:27
ppetrakiok19:27
stgraberso if the first command you run is "lxc profile show default" then it very likely will show you the old values19:27
ppetrakistgraber, I'll just poll for the right keys to show up then19:27
stgraberif it's the second command you run, it should be fine as the profile update script is very quick and starts immediately when the API is available19:27
sarnoldis there a do-nothing command that'll start it that would be a nice first command to run?19:28
stgraberppetraki: why do you need to wait?19:28
stgrabersarnold: lxc finger19:28
sarnoldstgraber: looks perfect :)19:28
ppetrakistgraber, I don't. Just like you said, if I run the lxc command again it should have reflect the correct profile19:29
sarnoldppetraki: try: 'lxc finger ; lxc profile show default' and see if that's the results you expect19:29
ppetrakistgraber, so when I script this, I'll just call that and wait until bridge0 actually shows up in the config19:29
stgraberppetraki: well, my question is more, why do you need to check that bridge0 made it in there?19:29
ppetrakistgraber, its correct19:29
ppetrakistgraber, about to create a bunch of containers with it via script19:30
stgraberppetraki: if the next thing you do is launch a container, lxd will already wait for profile config updates to be done before actually starting it19:30
ppetrakiso I want it to be correct19:30
ppetrakistgraber, ok, I will try that, less work, better19:30
stgraberyeah, we sure didn't want users running "lxc launch ubuntu:14.04 blah" as their first command and that picking up the wrong bridge, so we do have logic to avoid that very problem :)19:31
ppetrakiI'm sure. It's just that I just spent a hour or so working with a slightly older version that wasn't as well behaved. So I got a little paranoid19:32
stgraberwe just can't do the same with the profile API as the profile update script itself needs the API to be functional to update the config :)19:32
stgraberah yeah, that particular fix was introduced in rc919:32
ppetrakistgraber, good stuff so far, it looks like it'll suit us well. passing through kernel devices was fun19:32
stgraber:)19:33
ppetrakinow to get the rest of the stack up19:33
ppetrakistgraber, ugh, check this out. it's like the default profile isn't the default. http://pastebin.ubuntu.com/15696865/19:37
stgraberppetraki: nope, you're just not reading this right :)19:37
ppetrakiof course!19:37
stgraberppetraki: it's listing you the interfaces INSIDE the container19:37
stgraberppetraki: ubuntu:x/amd64 is pretty dated (last beta) so it still has lxcbr019:38
stgraberppetraki: if you were to use ubuntu-daily:x/amd64 you wouldn't see that interface anymore19:38
sdezielstgraber: nested lxc ?19:38
* ppetraki switches images19:38
stgrabersdeziel: yeah, it's showing the lxcbr0 that's running inside the container as we have lxd pre-installed in cloud images19:38
sdezielmakes sense, thanks19:39
sarnoldowwwww19:40
sarnoldnow my head hurts19:40
* patdk-wk is so far happy with lxc, though, it kindof annoying to adjust my apparmor profiles for it19:42
patdk-wkor is there a better way to *stack* apparmor profiles19:43
patdk-wkthink I use changehat :(19:43
patdk-wkah, used aa_change_profile19:44
tyhickspatdk-wk: the Xenial kernel has support for stacking AppArmor profiles19:45
sarnoldpatdk-wk: I -think- that just landed..19:45
patdk-wkya, I'm using 16.0419:46
sarnoldpatdk-wk: .. dunno if lxd has adapted yet19:46
tyhickspatdk-wk: the required userspace changes will hopefully land within the next few days19:46
tyhickssarnold: no, they haven't19:46
patdk-wkI found out my profile change didn't work, so I created a lxc profile + my changes19:46
patdk-wkthen adjust the lxc profile to allow that profile to be used19:46
patdk-wkworked for now, just seems like a bit of extra work19:46
patdk-wkok19:46
tyhickspatdk-wk: you'll (hopefully) soon have aa_stack_profile() in Xenial's libapparmor19:46
patdk-wkso that will only work when running lxc 16.04 guests?19:47
patdk-wkbut not 14.04 right?19:47
patdk-wkunless backport libapparmor somehow19:47
sarnoldthe stacks would be managed by the 16.04 lts system; i'd expect it to work..19:47
tyhickspatdk-wk: it'll require a 16.04 host19:48
tyhickspatdk-wk: note that lxc/lxd integration for stacked profiles is not in place yet19:48
patdk-wkoh19:48
tyhickspatdk-wk: as long as you have a 16.04 host, it should work with any ubuntu container, including 14.0419:49
patdk-wkso the inside will call normal aparmor call, and it will be adjusted to a stack call?19:49
patdk-wkI think that is what I don't get :)19:49
patdk-wkok :)19:49
tyhicksthe container setup code in the host will change to aa_stack_profile() (or simply adjust policy)19:49
tyhicksinside the container, nothing needs to change19:50
patdk-wkwell, I have a good test for it when it's ready atleast :)19:50
tyhicksgreat :)19:50
patdk-wkupdating my profiles for 16.04 was fun, so many new things I can limit :)19:50
tyhicks:)19:51
sarnold:D19:51
=== AndyTechGuy_ is now known as AndyTechGuy
ppetrakistgraber, I guess I can't call this from a container? mlockall( MCL_CURRENT|MCL_FUTURE ); our process doesn't want to be swapped out21:04
stgraberppetraki: I'm guessing that's a privileged call so not for unprivileged containers21:05
stgraberppetraki: you could try to make the container privileged though (lxc config set NAME security.privileged true && lxc restart NAME)21:06
ppetrakistgraber, can I make it a privledged container?21:06
stgraberat which point root in the container will be real root outside of it, so unless the kernel is being weird because of cgroups and stuff, this should work21:06
ppetrakistgraber, that's working21:09
ppetrakiit got further, and then I ran out of space :). good, it looks like it's getting the resources it needs21:10
Velus-universehello im folling this https://help.ubuntu.com/community/PostfixVirtualMailBoxClamSmtpHowto and i get to  this bit Make Dovecot listen for authentication requests and it says the code needs to be updated for newer version? can someone help me with this bit please?21:47
* patdk-lap wonders why it wouldn't work21:54
Velus-universehttps://help.ubuntu.com/community/PostfixVirtualMailBoxClamSmtpHowto#Configure_Dovecot shows a pre and post 12.4 setup and they look different, i cant figure out how it needs to be changesd21:56
Velus-universepatdk-lap, would you bhe able to help me with this please21:59
patdk-laphttp://wiki2.dovecot.org/HowTo/PostfixAndDovecotSASL22:00
patdk-lapthough, you still need the passdb and userdb sections22:01
Velus-universepatdk-lap, would the service auth be like this then http://pastebin.com/EMuMbmnN22:05
patdk-lapremove executable22:07
patdk-lapremove user22:07
patdk-lapand the example you are following uses auth-client instead of auth22:08
Velus-universeok22:08
patdk-lapdunno why they are having you use cram-md5, that is horrible22:09
Velus-universeok patdk-lap ifd i remove the user and that it reuses connection22:10
patdk-lapodd, I don't have it22:14
Velus-universei got it working now anyway, now i just need to ad mx records to my domain name but i forgot how to do them lol22:17
patdk-lapjust mx?22:18
patdk-lapif you plan to send email you should have ptr, spf, and maybe dkim22:18
Velus-universeyeah i will be doing them in a bit but i need to figure out how to do the mx i will be setting up the ptr and spf after the mx22:18
conrmahrCan you format a disk with a RAID array on UBNTU server?22:19
patdk-lapconrmahr, what does that even mean?22:19
conrmahrI know I'm an idiot22:20
conrmahrI took an 4TB WD Red Drive from a Synology DiskStation22:20
conrmahrand installed it on a computer runnning Ubuntu Server but it can't mount it.22:21
patdk-lapwell ya22:21
patdk-lapno one claims synology uses some standard raid/filesystem that ubuntu understands22:21
conrmahrCan I just erase and format it to ext4?22:21
patdk-lapI personally have no idea what they use22:21
patdk-lapsure22:21
conrmahryeah that's what i want to do22:22
sarnoldpatdk-lap: i'd always had the impression that it was dm-raid with ext4; iirc they even use ecryptfs for their encrypted storage22:22
patdk-lapdunno22:22
sarnoldbut their own 'magic raid" thing may be different, non-standard..22:22
conrmahrf-synology22:22
conrmahrim sick of it22:22
patdk-lapmost of them use mdadm/dm-raid22:22
conrmahri just want to run my own Ubuntu server22:22
patdk-lapsome of them, heh, don't22:22
patdk-lapI know mine uses raid-3d22:23
patdk-lapbelieve it was made by ibm22:25
Velus-universepatdk-lap, i have done the mx record now yay22:25
=== jgrimm is now known as jgrimm-afk
conrmahrCan anyone help me format a HDD on Ubuntu server with a raid array?23:51
naccconrmahr: you mean you want to use SW RAID with the partitions of a single HDD?23:52
conrmahri just remove everything from a HDD and start fresh with a clean ext4 partition23:53
naccconrmahr: ok, maybe just restate what you are trying to do?23:54
conrmahrYes I have a computer with a 60GB SSD with Ubuntu Server on it23:55
conrmahrI want to install a 4TB WD Red HDD as a secondary23:55
bekksAnd where is that raid array then?23:56
conrmahrbut that WD HDD is from a Synology NAS which was formated with a RAID array23:56
bekksSo you need to put a new primary partition table on it.23:56
bekksDo you use UEFI?23:56
conrmahryes23:56
conrmahri think i've tried but it says it's "in use"23:57
bekksYou've tried what?23:57
conrmahrwith 'parted"23:57
conrmahrnew primary partition23:57
bekksUEFI has no primary partitions.23:57
bekksyou are still trying to use legacy mode.23:58
conrmahri'm pretty sure my BIOS is UEFI23:59
conrmahrlet me check23:59

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!