[06:05] <Mead> is there a system requirements or minimum recomended specs for ubuntu-server?  I'm trying to figure how large of partition I should create for my install drive.
[06:08] <hateball> Mead: https://help.ubuntu.com/lts/serverguide/preparing-to-install.html
[06:08] <hateball> Mead: It all depends what services you intend to run on it
[06:10] <Mead> mainly just kvm and guest OS's
[06:11] <Mead> could it be installed on say an 8 or 16 gig flash drive?
[06:17] <lordievader> Good morning.
[06:18] <Mead> good evening
[06:20] <lordievader> o/
[12:51] <zetheroo> Hi all - I installed Ubuntu Server 16.04 onto a freshly created VM (Proxmox) using the minimal Ubuntu ISO. I then installed gnome-session-flashback and xorg, but I cannot get the desktop env to start with 'sudo startx' and it doesn't start on bootup.
[12:52] <zetheroo> my xorg log is as follows http://paste.ubuntu.com/24287725/
[12:55] <rbasak> zetheroo: server people don't know xorg. Try #ubuntu.
[12:55] <zetheroo> ok
[12:56] <rbasak> zetheroo: seems odd that you're using startx though. Usually people use lightdm or similar. Why not just use the desktop installer if that's wha tyou want?
[13:01] <blackflow> zetheroo: looks like there's no video device
[13:01] <blackflow> no dri, no framebuffer
[13:02] <zetheroo> aparantly I need a gdm\
[13:02] <zetheroo> no video device? hmm .. ok
[13:02] <blackflow> zetheroo: well the errror log you pasted shows no valid gpu, look at the (EE) entries
[13:03] <zetheroo> its a VM set to VMware compatible ... whatever that is :P
[13:03] <blackflow> could be just appropriate driver missing
[13:05] <blackflow> zetheroo: otoh, check if xserver-xorg-video-vmware is installed
[13:06] <zetheroo> installed lightdm and rebooted  - I have login screen
[13:06] <zetheroo> though the resolution is huge - LOL
[13:08] <zetheroo> ok, I changed the VM display settings back to default (VNC) and now have a more sensible resolution at login. BUt when trying to login I get 'Failed to start session' in red text above the password input area
[13:09] <blackflow> zetheroo: you'll have to check the logs (perhaps via ssh)
[13:10] <zetheroo> I am in there but not sure which logs to be hunting down
[13:10] <blackflow> zetheroo: but indeed like rbasak said, at this point the problems are likely related to non-servery parts, so better ask in #ubuntu
[13:10] <blackflow> zetheroo: journalctl -n -p err   is a good start
[13:10] <zetheroo> ok, I'll duck back in there then :D
[13:11] <zetheroo> whole bunch of this:
[13:11] <zetheroo> Mar 31 15:07:33 orbit1604 lightdm[1344]: PAM unable to dlopen(pam_kwallet5.so): /lib/security/pam_kwallet5.so: cannot
[13:11] <zetheroo> Mar 31 15:07:33 orbit1604 lightdm[1344]: PAM adding faulty module: pam_kwallet5.so
[13:12] <blackflow> kwallet is kde... did you install kde?
[13:13] <zetheroo> no, I think thats installed by default ... even on Ubuntu Desktop
[13:13] <blackflow> (I see you asked in #ubuntu too, so yeah, bettre there)
[13:15] <Ussat> FUK
[13:16] <zetheroo> 👍
[14:48] <jge> hey all, wondering if I can install php5.6 on Ubuntu 16 without adding the odrej/php PPA?
[14:48] <jge> I'm working with a server that has no internet connectivity and will never have one.
[14:49] <jge> It only has access to Ubuntu repos
[14:52] <rizonz> meh 8G for / seperated form the rest is not enough anymore
[14:52] <rizonz> *from
[14:53] <nacc> jge: no, you cannot
[14:57] <rbasak> jge: you could create a 14.04 container inside the 16.04 machine and have php 5.5.
[14:57] <rbasak> jge: or, if it'll never have internet connectivity, install Vivid from an old archive and never update it. It'll be full of security holes but perhaps that won't matter?
[14:58] <rbasak> Though if it won't matter, then ondrej's PPA shouldn't either I suppose, so perhaps I don't understand your scenario.
[15:01] <jge> rbasak: I just asked the firewall team, they agreed to allow access to that PPA (life's good again ;)
[15:03] <nacc> jge: not sure why you can't just use 14.04?
[16:09] <dn`> I try to boot an iscsi Ubuntu via ipxe; I was able to install Ubuntu on the ipxe target via; on reboot I always get a ‘Login error (Login status  0) on conn 7; 18 - encountered non-retryable iSCSI login failure; I disable any kind of authentication and was able to install on the target - anyone got an idea why it’s not booting from it?
[16:10] <nacc> dn`: does it drop you to a shell?
[16:10] <dn`> yes
[16:11] <nacc> dn`: are you able to login to the  iscsi target from the shell?
[16:11] <dn`> (it’s the first time I do this, so it might be something very stupid I’m missing)
[16:11] <dn`> … it just booted...
[16:11] <dn`> I think I know now why
[16:11] <dn`> one sec..
[16:12] <dn`> I think the ipxe(?) keept the connection open and I set the max con limit per target to 1
[16:12] <nacc> dn`: ah
[16:12] <dn`> just to debug I set it to 2 .. but that can’t be slution can it?
[16:13] <nacc> dn`: not sure
[16:14] <dn`> this took me longer then it should have;-)
[16:33] <dn`> nacc: maybe another question, my basic plan was to prepare one lun - then clone this lun for each machine I need to install; so for each machine I assume I need to adapt the config - so I greped for the IQN in /etc and found it a various places - the most prominent is e.g. iscsi.iscsi.initframfs that I need to adapt — do you know if the files in iscsi/nodes/iqn… or iscsi/send_targets/:IP… get auto generated?
[16:34] <nacc> dn`: you mean you were going to install using one lun and then treat that as a 'master image'?
[16:35] <dn`> yes - I thought that idea would be good - bad idea?
[16:35] <dn`> the problem is I would also do it with preseed/kickstart - I wouldn’t mind but I can’t find a iscsi example;-)
[16:36] <nacc> dn`: so the problem you'll find, depending on your config is
[16:36] <nacc> a) all of them will share an initiator name
[16:37] <nacc> b) you may want to read README.Debian from open-iscsi
[16:38] <dn`> I assumed I could change the initiator name for each - I don’t need to install them in parallel
[16:38] <nacc> dn`: i would suggest modifying your root imagea
[16:38] <nacc> dn`: and make it not used on its own
[16:38] <nacc> dn`: if that makes sense
[16:38] <dn`> not sure I follow - ‘not used on it’s own?’
[16:39] <nacc> dn`: some context on what i mean: https://naccblog.wordpress.com/2017/03/30/iscsi-initiator-names-in-cloud-images/
[16:39] <nacc> dn`: treat it as a master image
[16:39] <nacc> dn`: well, you can use it, i guess
[16:39] <nacc> dn`: but i mean, it sounds like basically you're doing your own little cloud :)
[16:39] <nacc> dn`: and that requires making sure your 'image' is clean
[16:39] <nacc> smoser: --^ :)
[16:39] <dn`> ah, yes - that’s kinda the plan
[16:39] <nacc> dn`: things like ssh host keys, iscsi initiator names, etc. should be unique
[16:40] <dn`> hmhm, ssh host keys is a good point - guess I’ll try to figure out how todo it with preseed/kickstart
[16:40] <dn`> but this will create many tears from past tries ;-)
[16:40] <nacc> dn`: we (server team) are talking about this generally
[16:41] <nacc> dn`: but i don't think it's 'solved' yet :)
[16:41] <dn`> iscsi via preseed?
[16:41] <nacc> dn`: making a clean image to use
[16:41] <nacc> dn`: given an installed instance
[16:41] <dn`> ah - I think it’s not insane hard
[16:41] <dn`> there is a command to regnerate the host key or you could just delete it (auto generate on start then)
[16:41] <nacc> without having tohave a script go and do it :)
[16:41] <dn`> ah;-)
[16:41] <dn`> yeap, that would be nice - but if I can have one request
[16:41] <dn`> “fix” pressed ;-)
[16:42] <dn`> I often have to install machines without ipmi
[16:42] <nacc> what's wrong with preseed? or do you mean iscsi support in preseed?
[16:42] <dn`> and then preseed can be a super awesome tool
[16:42] <dn`> yes, iscsi support and something different then partman ;-)
[16:43] <smoser> nacc, just saying this out loud, i think that the "updated" iscsi.initiatorname probalby gets put into the initramfs at some point, and at that point the initramfs has a copy and is also then dirty.
[16:43] <dn`> e.g. assuming you have a 100GB disk, and you want something like 30GB used for / [ignore th enumbers etc] - you are forced to also create 70g parition - otherwise partman fails; you can’t ‘not’ create it
[16:43] <nacc> dn`: tbh, there's an entirely new instller in development (subiquity) -- not sure on preseed support there
[16:43] <nacc> powersj: you asked about that on the list, right?
[16:44] <nacc> smoser: yep, i think that's true, as there is an initramfs hook in the package
[16:44] <dn`> nacc: can subiquity do a non interactive install?:)
[16:44] <dn`> how stable is it?:) *googleing*
[16:44] <nacc> dn`: no idea :)
[16:44] <nacc> dn`: not stable, afaict
[16:44] <nacc> dn`: it's still internal, mmaybe
[16:45] <dn`> found the repo
[16:45] <nacc> dn`: yeah, it is in active development
[16:45] <dn`> but I guess I’ll try to figure out if iscsi is or is not possible with preseed :/
[16:45] <nacc> this at least refers to it: https://gist.github.com/Lupul/0accbdc94b8a5ce674d7
[16:46] <powersj> nacc: yeah I did ask about it. I think still need to understand where it would be used and by whom
[16:46] <smoser> if you're doing iscsi root... then it really seems best to me to shoot for doing image based clone rather than install.
[16:46] <smoser> and use cloud-init or some other "first time" thing.
[16:47] <nacc> dn`: --^
[16:47] <smoser> basically then, get yourself a pristine image (from cloud-images.ubuntu.com, have to add a kernel if you need a hardware kernel)
[16:48] <smoser> and then just clone that thing. probably can even get away with putting a static /var/lib/cloud/seed/ in it for your metadata and user-data. or you can clone and then write those few files.
[16:49] <dn`> smoser: that’s what I tried, will look more into it - the problem is I need to clone the image at least twice I think for each installation ;-)
[16:49] <dn`> nacc: thanks - will try to figure that out
[16:49] <dn`> nacc: maybe it’s easy as addres sis the IP and multiselect is the name/iqn
[16:49] <dn`> that would be WOHA ;-)
[16:56] <drab_> how woudl you evaluate the different performances of more slower cores vs fewer faster cores when it comes to concurrency?
[16:57] <nacc> drab: what do you care about? define a performance metric and measure it
[16:58] <drab> nacc: yeah, the thing is, I don't have the hw test to on. So I'm not sure if 8 cores at 2.5GHz are going to do better than 4 cores at 3.2Ghz
[16:59] <nacc> drab: ... so how would you evalutate this at all? thought experiment?
[17:00] <drab> nacc: I was hoping someone here would have gone through similar considerations and maybe had the gear to test it, and knew that for eaxmple in a hosting scenario as a rule of thumb you're better off with faster CPUs and fewer core count or whatever configuration
[17:00] <drab> I coulsd possibly try to recreate something with containers and limiting cores, but don't think I can influence freq at all
[17:00] <nacc> drab: right, you never said 'hosting scenario' -- so # of VMs is a performance metric
[20:44] <arooni> what should i check when i get a 'Connection refused' message from my ubuntu 14.04 server when i try to ssh to it?
[20:44] <arooni> its being antisocial
[20:44] <dlloyd> is openssh running?
[20:48] <tomreyn> arooni: connection refused suggests that some endpoint or intermediary recived your connectrion and decided to actively drop it.
[20:49] <tomreyn> that's what an openssh server does when it's just starting up, or when you are connecting from an ip address not authorized to connect to it.
[20:49] <tomreyn> but also a firewall, wherther host or network based could do such.
[20:49] <tomreyn> most of the time, firewalls would just drop the traffic, though
[20:49] <tomreyn> (which would result in a timeout message, not the one you got)
[22:41] <Mead> welp I took the plunge, I've got ubuntu server installed,  nothing but command line,  now I have no choice but to learn how to configure everything in the CLI...   so...  now I guess I need a load of guides and documentation.  Anyone kwow were to start?  The more searchable the better.
[22:51] <blackflow> Mead: start with specific software you want/need
[22:56] <compdoc> you can also install a minimal desktop like mate. which then allows you to remote into it
[22:58] <compdoc> my life would be a living hell with remote dektops
[22:59] <blackflow> that kinda defeats the purpose :)
[23:02] <compdoc> nothing but net. I get it
[23:15] <bekks> Mead: you could even install a graphical environment. Which will not help much in configuring services on your server.