/srv/irclogs.ubuntu.com/2016/11/25/#ubuntu-server.txt

=== nchambers is now known as god
=== god is now known as nchambers
=== JanC_ is now known as JanC
=== amoralej|off is now known as amoralej
=== _degorenko|afk is now known as degorenko
jonahHi can anyone please help with TCP Wrappers? Are they supported in 16.04 LTS as standard or do I have to install a package? Also how can I check if an application is built to support them?11:30
jonahThanks for any help with this11:30
jonahIs there a way to add tcp-wrappers support into a package with apt or do you have to compile from source to get tcp wrappers?11:44
=== iberezovskiy|off is now known as iberezovskiy
=== amoralej is now known as amoralej|lunch
zioprotoIs it the thanksgiving weekend right ? :) I guess most people on holiday ?13:59
zioproto:)13:59
zioprotojamespage: looks like we have another chance to work together https://bugs.launchpad.net/nova/+bug/164483914:00
ubottuLaunchpad bug 1644839 in OpenStack Compute (nova) "If flavor has more than 32 cpus cannot spawn instance if glance image has hw_vif_multiqueue_enabled='true'" [Undecided,New]14:00
AlecTaylorhi14:03
AlecTaylorShould I place my `.sock`s in `/run` like everyone else, or should I create a new dir like some people, e.g. /softwarename/var/api/api.sock14:04
zioprotoAnyone knows how to check the value for MAX_TAP_QUEUES for Kernel 4.4.0-47-generic #68~14.04.1-Ubuntu SMP ?14:21
jamespagezioproto, hello!14:24
zioprotojamespage: hello :) I am dealing with a nice bug in production14:25
zioprotoas a workaround I told my use to upload a glance image with hw_vif_multiqueue_enabled='false'14:25
zioprotoso he can boot the instance14:25
zioprotobut he gets then the slow networking14:25
zioprotothe patch is easy, I can rework https://review.openstack.org/#/c/332660/14:26
=== amoralej|lunch is now known as amoralej
zioprotocloning git clone git://kernel.ubuntu.com/ubuntu/ubuntu-trusty.git to see if MAX_TAP_QUEUES changes from vanilla kernel14:28
zioprotojamespage: the ubuntu kernel is different14:36
zioproto#define MAX_TAP_QUEUES DEFAULT_MAX_NUM_RSS_QUEUES14:36
zioprotothis brakes nova14:36
zioproto:(14:36
jamespagezioproto, trusty is 3.13 right?14:37
jamespageoh wait I see you're using the 4.4. kernel14:37
zioprotoyes14:39
zioprotoso there is this patch in nova that hardcodes some values14:39
zioprotothere is also nice reference table of this values, with URL to the git of vanilla kernel14:39
zioprotojamespage: look at this review https://review.openstack.org/#/c/332660/14:40
zioprotoI am runnig this kernel linux-generic-lts-xenial             4.4.0.47.3414:41
jamespagezioproto, xenial kernel - http://paste.ubuntu.com/23532517/14:43
jamespageshould be defaulting to 25614:43
fricklerjamespage: seems that value is changed when building the kernel for trusty14:44
jamespagehmm14:45
zioprotoLinux zhdk0088 4.4.0-47-generic14:45
zioprotohow can I check it at runtime ?14:45
zioprotofrickler: how you find that out ?14:46
jamespagezioproto, I see the same stanza in drivers/net/tun.c in the xenial HWE branch14:49
fricklerzioproto: ah, no, sorry, I was looking at the wrong branch14:49
zioprotoDid you guys see my stacktrace  in the bug report ?14:50
fricklerso according to the comment in baf71c5c1f80d82e92924050a60b5baaf97e3094, this should be good for 256 vCPUs14:50
zioprotobut I get14:51
zioprotoInvalid number of queues (= 46), must be a postive integer less than 31.14:51
fricklerzioproto: just to make sure, you do run on a system with > 32 cpus?14:52
zioprotohttp://paste.ubuntu.com/23532554/14:53
zioprotoyes because If I remove the multiqueue property from the glance image I am able to boot the instance14:53
zioprotoI have 48 CPUs on the sistem14:54
zioprotoand now the instance is running14:54
zioprotoI told the user14:54
jamespagezioproto, which libvirt version are you using?14:55
zioprototo boot from a different glance image where I decorated the image with hw_vif_multiqueue_enabled='false'14:55
zioprotojamespage: 1.2.16-2ubuntu11.15.10.4~cloud014:55
fricklerzioproto: so as I workaround maybe try to set the limit in the nova code you referenced to 31 instead of 25614:55
jamespageactually that message comes from qemu14:55
zioprotofrickler: yes that would work but I dont want to carry local patches14:55
zioprotofrickler: I need to come up with a solution that will be accepted in openstack upstream14:56
zioprotojamespage: qemu 1:2.3+dfsg-5ubuntu9.4~cloud214:56
jamespagegive me a sec14:58
jamespagezioproto, yup you're being blocked by qemu14:59
fricklerzioproto: maybe the limit 31 31 is imposed by qemu, so nova is wrong in using that kernel value.14:59
jamespageVIRTIO_PCI_QUEUE_MAX == 64 in the version you're using15:00
jamespage64 /2 -115:00
jamespage(VIRTIO_PCI_QUEUE_MAX - 1) / 215:00
zioprotojamespage: you rock ! how did you find that out ?15:00
cpaelzerit is surely arch specific, I've had s390 on 64 cpus, but then no PCI there15:00
jamespagezioproto, been poking at qemu today - but 2.715:00
jamespagewhere its 102415:00
zioprotoI need to update this bug: https://bugs.launchpad.net/nova/+bug/157063115:01
jamespagecpaelzer, that constant has a new name15:01
ubottuLaunchpad bug 1570631 in nova (Ubuntu) "With hw:vif_multiqueue_enabled, libvirt driver fails with VM larger than 8 vCPU" [Low,Fix released]15:01
jamespageVIRTIO_QUEUE_MAX15:01
jamespagenot PIC15:01
jamespagePCI rather15:01
jamespageinclude/hw/virtio/virtio.h:#define VIRTIO_QUEUE_MAX 102415:01
jamespagethats from 2.7.015:01
cpaelzerthanks jamespage15:01
jamespagecpaelzer, yw15:01
cpaelzerI just say it is arch specific as I had 64 running already15:01
jamespagecpaelzer, I'm putting qemu in a snap with libvirt15:02
cpaelzerah thanks, just wondered what you do with that atm15:02
jamespagezioproto, xenial qemu has:15:03
jamespageVIRTIO_QUEUE_MAX15:03
jamespage=include/hw/virtio/virtio.h:#define VIRTIO_QUEUE_MAX 102415:03
jamespagezioproto, so this looks like an issue isolated to the liberty UCA version you're using15:04
fricklerright, that would mean that nova has to check the qemu version in addition to the kernel version and set its limit accordingly, I guess15:04
cpaelzerzioproto: but x86 HW virt is limited to stay <=HW-cpus IIRC - so you can go to 48, but not test e.g. 256 (maybe that can be tuned thou)15:04
fricklerjamespage: MIN_QEMU_VERSION in nova is still at 1.5.315:04
zioprotojamespage:  should we make a patch just for the ubuntu packages ?15:04
jamespagefrickler, youch15:04
jamespagezioproto, for qemu15:04
jamespage?15:04
jamespagezioproto, or for nova?15:05
fricklerjamespage: may be bumped to 2.1.0 for pike iiuc15:05
jamespage\o/15:05
jamespagethat's still not enough for 102415:05
rbasakcpaelzer: around? Can I give you an intermediate report on this strongswan merge review? I'd like to talk it over with you before the complexity of our delta makes me lose my mind :)15:05
jamespagerbasak, oh I remember that one15:05
rbasakcpaelzer: http://paste.ubuntu.com/23532604/ are my current notes.15:06
cpaelzerrbasak: I'll get a visitor to teach math to soon, but til then ok15:06
rbasakcpaelzer: hangout? Two minutes.15:06
cpaelzerrbasak: give me two minutes to clean up the next evil merge15:06
cpaelzerrbasak: ok15:06
zioprotojamespage: I guess for nova15:06
fricklerzioproto: if you don't come up with something else, I'll propose a patch to nova upstream next week15:06
rbasakSnap :)15:06
fricklergotta run for the weekend now15:07
zioprotoI am also on my way to the weekend15:07
zioprotolets meet again here on monday :)15:07
rbasakcpaelzer: https://hangouts.google.com/hangouts/_/canonical.com/strongswan?hl=en-GB&authuser=015:07
jamespagezioproto, frickler: have a nice one guys15:08
zioprotobye !!15:35
=== degorenko is now known as _degorenko|afk
=== iberezovskiy is now known as iberezovskiy|off
hheeguys, i have got a server with root enabled by default. how can i turn it off again?17:52
geniihhee: First, make sure you have a user with sudo rights so you don't lock yourself out of the system. If all good there, sudo usermod -L root    ...will lock the root again18:02
geniihhee: To make sure it took: sudo grep root /etc/shadow   ...and make sure the second field is an exclamation mark, like:  root:!:18:04
hheegenii, got it. already create usual user18:06
hheegenii, thx a lot18:06
hhee:)18:06
geniiGlad to assist18:09
hheegenii, well another question18:15
hheeseems like i didn't have apropriate record in sudoers file18:15
hheewhich way more eight to add user with sudo right?18:16
hheerights*18:16
hheegot it18:21
hhee!18:21
tarpmanhhee: the usual way in ubuntu/debian is to add the user to the 'sudo' group18:22
hheetarpman, got it, thx :)18:22
tarpmanhhee: usermod -aG sudo username18:22
tarpmanok18:22
=== amoralej is now known as amoralej|off
hheeguys. question.18:28
hheedoes ubuntu-server has groups for install software? how can i do it?18:28
hheefor example LAMP and mail server18:28
hheeim reading man apt, but dont see smth similar18:30
tarpmanhhee: yes. have a look at tasksel18:30
hheetarpman, got it. very handy.18:32
hheeone more question. after many command in ssh session i have got something like this - perl: warning: Setting locale failed.18:33
hheeperl: warning: Please check that your locale settings:18:33
hheeafter locale command18:34
hheegot - locale: Cannot set LC_CTYPE to default locale: No such file or directory18:34
hheewhat i did wrong?18:35
tarpmanwhat's your locale set to?18:35
hheetarpman, i thought it was utf818:35
tarpman(run the locale command with no arguments to get a summary)18:35
hheetarpman, LANG=en_US.UTF-818:35
hheenot set - LANGUAGE= and LANGUAGE=18:35
hheesorry and LC_ALL=18:36
tarpmanhhee: do you have language-pack-en installed18:37
hheetarpman, cool. got it. i have found http://askubuntu.com/questions/162391/how-do-i-fix-my-locale-issue18:39
hheetarpman, and there is solutions18:39
tarpmanok18:39
hheetarpman, big tnx for awesome support.18:39
hheehave a nice day!18:39
tarpmannot sure those are the best solutions, but if they work, fine :)18:39
hheeseems like work :)18:39
hheeguys. another quesion about sudo :)19:06
hheei tried - sudo cat file.conf > file2.conf19:07
hheebash: file2.conf: Permission denied19:07
hheewhich way to do it?19:07
tarpmanyes, because 'cat file.conf' runs under sudo, but '> file2.conf' happens in your own shell, outside of sudo19:07
tarpmanyou want: sudo tee file2.conf < file.conf19:08
tarpmanwhich reads from file.conf in your current shell and writes to file2.conf under sudo19:08
hheetarpman, cool... thx :)19:08
hheetarpman, another question19:27
hheesudo cd /var/log/apache2/19:27
hheesudo: cd: command not found19:28
hheehow can i handle this? :)19:28
tarpmanhhee: if "cd" did exist as a standalone command, what would it do?19:28
tarpmanhhee: your shell would spawn a "sudo" process, which would execute "cd", and then exit19:28
tarpmanhhee: meanwhile your shell is exactly where it was, nothing's changed19:28
hheetarpman, i need some group of command.. i have to think19:29
tarpmanhhee: I don't know what you're trying to do. maybe you want to cd /var/log/apache2 and then sudo something in there. or maybe you want to open a root shell (sudo -s or sudo -i) and do stuff in that shell19:29
hheetarpman, got it. sudo -i what i need19:32
hheetarpman, thx again :) many times19:32
=== binia is now known as binia_the
=== binia_the is now known as binia

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