[02:00] <TAFB> can anyone help me install Windows XP on my Ubuntu VPS?
[02:03] <sarnold> TAFB: heh are you still trying to use that openvz-ish system? if so, start with "bochs"
[02:05] <TAFB> sarnold: Yes, I believe so :)
[02:05] <TAFB> <- linux noob
[02:06] <TAFB> bochs looks interesting, I'll read up on it, thanks!
[02:10] <sarnold> TAFB: another thing to consider is just paying amazon to run windows for you -- https://aws.amazon.com/marketplace/pp/B00KQOWCAQ/ref=mkt_ste_windows_amis -- I haven't got a clue how you -use- the thing, but maybe there's a VNC interface or something?
[02:11] <RoyK> TAFB: probably easier with a kvm vm thnan openvz
[02:11] <TAFB> I'd really really like to run it on my current server as that's where all the media will be streamed from, but if it's not possible or has bad reliablility or performance then I'll consider other options.
[02:11] <sarnold> a t2.medium has 2vcpu and 4gigs ram, and costs just under 8 cents an hour :) that buys a lot of not-caring-how-it-works :)
[02:11] <TAFB> I have a Windows RDP server in Paris but the peering/speed to north america is horrible.
[02:17] <RoyK> TAFB: just setup samba and share the media that way
[02:18] <TAFB> it's for a live stream, so latency and jitter is an issue
[02:19] <RoyK> then why windows?
[02:19] <RoyK> or why mixed platforms?
[02:19] <TAFB> the only software I've found to unencrypt the live stream and re-stream it is tubedigger, only runs on windows.
[02:19] <sarnold> hah then forget you ever heard about bochs :)
[02:20] <TAFB> bochs not good performance? lol
[02:20] <TAFB> it's windows xp, how hard can it be to emulate?
[02:21] <RoyK> TAFB: not very hard at least if you install the paravirt drivers
[02:21] <RoyK> TAFB: if you don't paravirt it, it'll be very slow
[02:26] <TAFB> "Paravirtualization requires the guest operating system to be explicitly ported for the para-API"
[02:26] <TAFB> doesn't sound like I'll be able to use it on my current VPS :(
[02:27] <sarnold> indeed no :)
[02:57] <patdk-lap> I think everyone has discontinued paravirtualization
[02:57] <patdk-lap> except if your running like lxc :)
[02:57] <patdk-lap> I know vmware killed it off, and aws did, not sure if xen did or not
[02:58] <sarnold> paravirtualization has moved into the drivers
[02:59] <sarnold> virtio block, virtio nic, virtio rng, etc
[02:59] <RoyK> patdk-lap: hwvirt is rather dead - as sarnold says, paravirt is in the drivers
[03:00] <RoyK> patdk-lap: xen is rather dead, though, which is good
[03:00] <sarnold> unless you're still running with an ne2k emulated nic :)
[03:01] <RoyK> heh - I setup this vm a few days back to feed flightradar24 with an usb thing - turned out it was using a crappy usb2 virt driver with only polling
[03:02] <RoyK> dma was envented - what - 30 years ago?
[03:02] <patdk-lap> oh, not paravirtualization, paravirt drivers
[03:03] <RoyK> same thing
[03:03] <patdk-lap> vmware and aws killed paravirt vm's though
[03:03] <patdk-lap> no it's not
[03:03] <patdk-lap> you can run xen in paravirt or hvm mode
[03:03] <patdk-lap> paravirt shares the host os kernel
[03:03] <RoyK> yeah and the hvm mode sucks rather badly
[03:03] <patdk-lap> why? hvm is just normal qemu
[03:03] <patdk-lap> qemu-kvm
[03:04] <RoyK> xen in hwvirt
[03:04] <RoyK> sucks
[03:04] <patdk-lap> if your on 32bit, and not using cpu vt, sure
[03:05] <RoyK> performance of a 386 with old drives
[03:27] <TAFB> how do I set the resolution of tightvnc? I tried xrandr --fb 1920x1080 but it says "can't open display"
[06:47] <rostam> hi all using 14.04 server trying to resolve the boot issue, I found out the plymouth restart fails, I changed the grub to disable it still I see the issue, How can I disable the plymouth? Thank you.
[07:15] <temhaa> Hello, I am newbie in virtualization. I am thinking to use ubuntu as host server. Which virtualization method is best. I will buy tower server for this. For example I need 8 GB RAM, 2 or 4 core CPU for each guest. I want to create least 3 or 4 guest for one host.
[07:16] <hateball> !best
[07:16] <hateball> ugh, spring cleaning
[07:16] <temhaa> Actually I need to 4 machine. I am thinking to provide with virtualization. What are your advices. If you help me I would be glad
[07:16] <hateball> temhaa: depends what you are comfortable with using
[07:16] <hateball> !kvm
[07:18] <temhaa> hateball: I have java application, ELK stack, some little applications in digitalocean. I am thinking to move my psysical server.
[07:19] <temhaa> hateball: I tried kvm before. But it is best tool I am not sure. I cant use lxc because I have docker containers for these applications
[07:20] <temhaa> hateball: I wonder, If I open 4 guest with kvm (each guest must have 8 gb ram 4 core cpu). what must be resources of host machine? How can I caclulate. I couldnt learn this
[07:22] <temhaa> Is there anybody who experienced with kvm
[07:31] <hateball> temhaa: overprovisioning isnt very easy to calculate
[08:06] <cpaelzer> temhaa: https://www.stgraber.org/2016/04/13/lxd-2-0-docker-in-lxd-712/
[08:06] <cpaelzer> temhaa: docker is no "you can't use containers for the first level"
[08:07] <cpaelzer> temhaa: I really agree with hateball that the biggest factor is what you feel comfortable with
[08:07] <cpaelzer> temhaa: that said - nobody else can decide lxd vs kvm for you :-/
[08:08] <cpaelzer> temhaa: I've used kvm and lxd a lot, and they just have different specialties and use cases
[08:08] <cpaelzer> temhaa: for lxd I recommend reading the full series about lxd that I linked one piece of
[08:09] <cpaelzer> temhaa: after that you should have a feeling for that, for KVM there are a million docs out there - it depends on what you want to know
[08:10] <cpaelzer> temhaa: and for your ressource question it depends - on how muhc of that size is active memory and how busy the cpus are
[08:10] <cpaelzer> temhaa: in general the answer to any performance and/or sizing question is "it depends"
[08:12] <temhaa> cpaelzer: non-shared resources are very importand me. as I know lxc is using shared resource, am I right?
[08:14] <cpaelzer> temhaa: kvm shares it's cpus as well if you don't stop it from doing so
[08:14] <cpaelzer> temhaa: so you want to hard-dedicate the ressources?
[08:14] <cpaelzer> temhaa: more like partitioning than sharing ?
[08:15] <cpaelzer> temhaa: that LXD can do as well in some way, see https://www.stgraber.org/2016/03/26/lxd-2-0-resource-control-412/
[08:16] <temhaa> cpaelzer: There are a lot of things what I dont know. For example overcommiting. Just think. I have two application. Each application wants 8 gb ram and 4 core. I want to install each application to different guest.
[08:17] <cpaelzer> temhaa: overcommiting just means for that case that if you have 4x8GB "guests" your Host doesn't have 32G
[08:18] <cpaelzer> temhaa: usually lxd will be a bit better at that, as it doesn't have the issue that the guest page cache appears as non-volatile memory
[08:18] <cpaelzer> temhaa: in KVM the host (in the worst case) will swap guest page cache (which the guest could just throw away)
[08:19] <cpaelzer> temhaa: if anythin (IMHO) the decision between KVM and LXD is more-isolation (kvm) vs. more effeciency (LXD)
[08:19] <temhaa> cpaelzer: Yes. I am learning what is mean.  I need 8X2 gb ram for two application, If I buy 16gb host machine then is it good practice I am not sure.
[08:19] <temhaa> cpaelzer: Because I want to decrease cost.
[08:20] <cpaelzer> temhaa: if you can size whatever you want, buy not only sum-of-guests, but sum-of-guests+safety (for the Host)
[08:21] <cpaelzer> temhaa: if your target is to lower cost you have to reverse that and that implies you strive for more effeciencey which is usually where LXD wins
[08:21] <cpaelzer> and when you say 8x2GB - is that 8x2GB fully used?
[08:21] <cpaelzer> like 2GB JVM Heap?
[08:21] <temhaa> cpaelzer: by the wy I have heard LXD for now. It is something like lxc?
[08:21] <cpaelzer> because then your system needs at least a bit more around to breathe
[08:21] <temhaa> cpaelzer: they can be use fully
[08:22] <cpaelzer> temhaa: https://linuxcontainers.org/lxd/
[08:22] <temhaa> cpaelzer: yeah I am reading at the same time
[08:22] <cpaelzer> just take the hour or two and read  this and the artictle series I linked before
[08:24] <temhaa> cpaelzer: You say, "you can install docker to LXD. You can open guest machines with low cost". Because LXD is more lightweight then KVM
[08:24] <cpaelzer> temhaa: if you like more of a "show" http://www.grabber.xyz/2015/05/lxd-vs-kvm.r_7UbDgExdI.html
[08:24] <temhaa> cpaelzer: Last link what you sent doesnt work, it is not opening
[08:25] <cpaelzer> temhaa: that is true as long as your docker images are not breaking on the extra confinement that lxd adds
[08:25] <cpaelzer> hmm, let me check that link
[08:25] <cpaelzer> temhaa: https://www.youtube.com/watch?v=90oxad2r8_E better?
[08:26] <cpaelzer> temhaa: I'd say check the vid, read the LXD article series by stgraber and then decide
[08:26] <cpaelzer> temhaa: I'd even say give one of your application containers a test e.g. on your laptop inside lxd
[08:28] <temhaa> cpaelzer: You can understand me, I am not professional. I dont have big server. actually I dont have server. I am thinking to buy to leave from digitalocean and amazon. So I am looking for virtuallization tool and I need to choice true resource of host machine. :)
[08:28] <temhaa> cpaelzer: Yes I am listening that video
[08:29] <cpaelzer> temhaa: I can understand you, but I can't (and I'm convinced nobody can) decide it "for you
[08:29] <cpaelzer> temhaa: in fact I think sales people can decide for you (that is what they do) :-P
[08:30] <temhaa> cpaelzer: I know, so I am asking what is your advice:)). If I have more money, It doesnt matter :P
[08:30] <temhaa> cpaelzer: sorry my english is not very well
[08:30] <cpaelzer> it is fine, never mind
[08:31] <cpaelzer> temhaa: I (personal opinion) would experiment with LXD and if it is working go with that as it is more efficient and you listed price as a target. But then (usually) on the clouds where you already are you are usually cheaper, just not as consistent. I'd chose KVM if (even more) isolation is a must for you.
[08:32] <cpaelzer> temhaa: and without any offense - but with "I am not a professional" you are likely better off sticking with the clouds - as they are professionals maintaining that stuff for you
[08:33] <newcomer25> The whole Law is fulfilled in one statement: ‘You’ll love your neighbour as much as yourself’ - Galatians 5:14
[08:33] <newcomer25> God bless you all and have fun using chatting!
[08:33] <cpaelzer> that was weird
[08:36] <temhaa> cpaelzer: I dont need high avaliability.  I want to run application in true resource. I can do server management
[08:36] <temhaa> cpaelzer: I need to decide LXD or KVM. Therefore I need to read more documentation. your explainations are very helpfull for me
[10:42] <YamakasY> hi guys
[10:43] <YamakasY> when I ssh to an ubuntu box, which always worked, and I get a connection refused now (several servers) should that be logged in syslog ?
[10:45] <ogra> YamakasY, perhaps http://www.openssh.com/legacy.html ?
[11:18] <xmj> moin
[11:18] <xmj> where does 16.04'
[11:18] <xmj> 's postgresql server store its data?
[11:20] <maswan> same place as always, /var/lib/postgresql/ ... ?
[11:22] <xmj> thx
[11:38] <xmj> more on that note, which package contains that default data?
[11:58] <Odd_Bloke> xmj: What do you mean by "default data"?  AIUI, Postgres will create the files on install/first start.
[12:11] <xmj> all good
[12:11] <xmj> Odd_Bloke: when you intentionally remove the data in /var/lib/postgresql, the question is which package makes it reappear on reinstall
[12:11] <xmj> i eventually found that purging and fiddling around helps with ^
[13:08] <jamespage> coreycb, niggle with your ubuntu-reports change is that reports.qa.ubuntu.com can't access github - its also running 12.04
[13:08] <jamespage> urgh
[13:26] <coreycb>  jamespage, that's too bad.  12.04 might be ok but yeah it needs access to github.
[13:40] <GMAzrael> almight IRC, I have an issue. My 16.04.01 server is not reponsing to the local subnet, but is responding to NAT'd requests. UFW and iptables are off. The nic is bridged for containers. Any thoughts?
[13:40] <GMAzrael> responding to ssh attempts*
[13:44] <GMAzrael> tcpdumps say no packets are coming in, but the server is pinging fine
[13:44] <coreycb> jamespage, qemu 1:2.2+dfsg-5expubuntu9.7~cloud6 is ready to promote to kilo-proposed
[13:45] <jamespage> coreycb, looking now
[13:51] <jamespage> ddellav, hey - I did the MIR for oslo.privsep - could you look at the two new deps for heat:
[13:51] <jamespage> http://people.canonical.com/~ubuntu-archive/component-mismatches-proposed.svg
[13:51] <jamespage> for reference
[13:58] <coreycb> beisner, can you promote ceilometer 1:5.0.4-0ubuntu1~cloud0 ito liberty-proposed?
[13:59] <beisner> hi coreycb, done re: https://bugs.launchpad.net/bugs/1614570
[14:01] <coreycb> beisner, thanks
[14:02] <setuid> Something weird, if I install 10 kernels to do a regression, grub/update-grub shows those kernels, but only the most-recent one gets an entry in the visible config.
[14:03] <setuid> How can I get grub to see (and list) ALL of the installed kernels?
[14:03] <setuid> I'd rather not drop to the grub shell every time to find/choose/edit the kernel line to select the one I need to test
[14:04] <setuid> I installed grub-customizer, and IT sees all of them and creates entries, but then when I save and exit, they're reverted, and only the last one is set up in the default entry
[14:14] <xnox> setuid, all kernels should be listed under sub-menu.
[14:14] <xnox> no idea what grub-customizer is
[14:16] <beisner> yw coreycb
[14:16] <setuid> xnox, Agreed, and I can see all the entries in grub.cfg, but the UI doesn't show them
[14:17] <setuid> I'll cut down the number of kernels and see if maybe there's a limit here
[14:21] <trippeh_> setuid: it should show up under the "Advanced options for Ubuntu" grub sub menu during boot.
[14:22] <trippeh_> you have to select the menu for the list to show up.
[14:22] <setuid> Right, and that menu only shows 2 kernels; the default and safe-mode
[14:22] <trippeh_> ah ok, that is odd.
[14:23] <setuid> I'm going to wipe out all the kernels one by one and start again
[15:04] <setuid> trippeh_, confirmed, even with 5, 6, 10 kernels, it only ever shows the last one
[15:04] <setuid> this is 16.04 server
[15:05] <setuid> but grub.cfg is correct, and lists all entries
[15:06] <setuid> The output from update-grub also shows the correct entries
[15:08] <setuid> It's almost like it's just not getting written to the actual boot block
[15:10] <setuid> Hrm, interesting... the kernels are all -lowlatency kernels, but the ONLY entry that shows up in grub's menus, is the 4.2.0-27 *NON* lowlatency kernel, all the ones that are newer/later, do not show,and they're all lowlatency
[15:10] <setuid> parsing bug with the kernel filename? Or is there some host-based triggers that grub ignores?
[15:17] <setuid> just installed a handful of 4.x.y-generic kernels, they're all ignored by grub
[15:17] <setuid> it only shows 4.2.0-27
[15:17] <setuid> http://paste.ubuntu.com/23089276/
[15:20] <setuid> Just removed ALL the kernels, except 4.2.0-42-generic, and grub still shows 4.2.0-27 under Advanced, but the default entry boots 4.2.0-27 from -somewhere-, even though it's not in /boot/
[15:23] <nacc> setuid: i assume you're just reading /boot/grub/grub.cfg, right?
[15:24] <nacc> setuid: did this happen before you installed 'grub-customizer'?
[15:24] <setuid> nacc, Checked all the grub files, as well as the ncurses boot menu that comes up from the boot block when the machine is booted
[15:24] <setuid> nacc, This happened before I installed that, yes. The reason I installed it, was to see if grub-customizer correctly parsed the entries (it did)
[15:26] <nacc> setuid: i meant that the /boot/grub/grub.cfg file is also in error, not just the menu, sounds like it?
[15:26] <setuid> nacc, grub.cfg looks syntactically correct
[15:26] <nacc> setuid: it seems odd that 4.2.0-27 exists in teh menu and successfully boots if you removed it
[15:27] <nacc> setuid: i mean, does it contain menu entries for the other kernels?
[15:27] <setuid> # uname -a && grep 27 /boot/grub/grub.cfg
[15:27] <setuid> Linux ubuntu-14-04-4-vm 4.2.0-27-generic #32~14.04.1-Ubuntu SMP Fri Jan 22 15:32:26 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux
[15:27] <setuid>     insmod ieee1275_fb
[15:27] <setuid> i
[15:27] <setuid> I'm booted to a kernel that does not exist on the system
[15:27] <setuid> unless they're no longer stored in /boot/
[15:28] <nacc> setuid: are you on 14.04 or 16.04?
[15:28] <setuid> 14.04, (I misspoke earlier)
[15:29] <setuid> this is a clean, 14.04 image installed in vmware workstation
[15:29] <nacc> setuid: can you pastebin /boot/grub/grub.cfg?
[15:30] <setuid> Sure, one sec
[15:30] <nacc> fwiw, i think you're on a no-longer-supported HWE stack (14.04.4 rather than 14.04.5)
[15:31] <setuid> http://paste.ubuntu.com/23089322/
[15:31] <setuid> nacc, I used the 14.04.4 ISO to install it
[15:31] <setuid> but it shouldn't be anything magical, 14.04.4 works elsewhere
[15:32] <setuid> where is 4.2.0-27 actually booting from? It's nowhere on disk, nor listed in the package listing
[15:34] <setuid> ah-HA!
[15:34] <nacc> setuid: you found it?
[15:34] <setuid> So /boot/ was a 200MB partition, I moved that out of the way, created /boot/ off of the root, put my kernels in there, grub and all tools recognized it, so I just mounted /boot2/ (the original /boot/), and there were a number of kernels in there, including -27
[15:35] <nacc> ah
[15:35] <setuid> So I removed everything, and rebooted. Now it doesn't boot (as expected)
[15:35] <setuid> so grub-install probably was pointed to that partition, while userspace tools pointed to /boot/
[15:35] <setuid> hence, the schism
[15:36] <nacc> yep, changing /boot makes sense
[15:36] <setuid> I needed the space for testing dozens of kernels
[15:37] <nacc> setuid: yep, understood
[15:46] <setuid> nacc, chicken and egg... /boot/ now resides on / which is an LVM, so grub can't see/boot it
[15:46] <rattking> Hi folks, I am having an issue where my post-up rules for adding a static routes in networking/interfaces are not taking on boot, if I restart networking service the routes do take, if I add a sleep 1; to the post-up it does take as well, this same config worked with 12.04. it has eth0,1 bonding and vlans so its a bit more then a usual interfaces config.. any ideas on how I can debug that further?
[15:46] <rattking> thanks.
[15:47] <setuid> I'll need to reinstall the whole box, choose non-LVM, or create a fatter /boot and do manual partitioning
[15:47] <nacc> setuid: ah ... yeah it might be easier
[15:48] <setuid> Can't easily expand /boot/, so I'll just create a second disk for /boot/ and if I need to increase it, add a third disk, rsync, and mount it back (it iS a vm after all, easy to do)
[15:50] <nacc> setuid: yeah, undersizing boot has been, ime, the most commen reason t reinstall :)
[15:53] <setuid> It's not common to need to add many kernels, but 200MB is pretty small, esp. given the size of most drives these days
[15:54] <setuid> 20GB boot should give me enough breathing room
[15:54] <nacc> setuid: yeah, 200M is very small
[15:57] <setuid> 4 kernls and you're out of space
[15:58] <setuid> That was a bit tricky to find, given that I no longer had the old /boot/, but grub still installed itself with the old data.
[15:59] <setuid> Still though, update-grub had the right kernels, what was it doing, if not replacing grub on the boot record?
[15:59] <setuid> was it writing to the boot record on the LVM partition, which grub (at boot time) ignores?
[15:59] <setuid> because that would be a bug, update-grub should warn that it's doing that, and the system will not boot with the options used
[16:20] <rattking> it seems like my post-up rules are running before the network interface is actually up
[16:21] <setuid> rattking, how many cpus?
[16:21] <setuid> I know it seems unrelated, but there's an issue we tripped on recently that sounds similar
[16:21] <rattking> 4
[16:22] <rattking> but some servers will have more
[16:22] <setuid> humor me, boot one with 'maxcpus=4' and see what happens
[16:23] <patdk-wk> physical or cores?
[16:24] <rattking> those are cores.. no hyperthreading
[16:25] <patdk-wk> more I was wondering if the issue is related per core, or socket count
[16:25] <rattking> this motherboard is a dual socket, but only 1 is populated
[16:27] <rattking> wow.. with maxcpus=4 the postup rules loaded on boot.. what the heck kind of bug is that?
[16:29] <nacc> sounds like a race?
[16:29] <rattking> yeah I was thinking its a race.. because adding 'post-up sleep 1' caused it all to work too
[16:30] <nacc> rattking: is 4 < the number of cpus normally shown in /proc/cpuinfo ?
[16:31] <rattking> on this system its processor: 0 - 3 .. so yeah 4
[16:32] <nacc> setuid: hrm, do you know why specifying maxcpus= changed behavior in your system (why you suggested it)
[16:34] <setuid> nacc, we're tracking it down, it's some very subtle race condition with interfaces and sub-interfaces
[16:34] <setuid> "We" == (@Canonical)
[16:35] <setuid> There appears to be a tipping point somewhere north of maxcpus=8
[16:40] <rattking> setuid any ideas on other ways a can get around that? I have a bunch of systems with different numbers of cpus
[16:41] <setuid> Did it work when you tested it with statically defining maxcpus?
[16:43] <iliv> I have this odd problem with Xenial LTS running as guest vm in virtual box. It has two NIC's and one of them doesn't work after the system starts unless I restart networking or run ifdown enp0s8; ifup enp0s8 manually. ip a l command shows that link of enp0s8 is UP, it is assigned a static IP address but I can't ping other hosts on the same network unless I restart networking or toggle the enp0s8 interface manually with ifdown/ifup commands.
[16:43] <iliv> this is my dmesg log https://dpaste.de/7r2D/raw
[16:44] <setuid> iliv, Does /etc/network/interfaces have that interface set to start?
[16:44] <iliv> yes
[16:44] <setuid> auto enp0s8?
[16:44] <setuid> and a block for that adapter to define it? (iface enp0s8 inet static... etc.?)
[16:44] <setuid> or dhcp, as your needs require
[16:44] <rattking> setuid: I am not sure what you mean by statically defining maxcpus
[16:45] <setuid> rattking, boot your server, passing 'maxcpus=4' on the kernel args line ('e' at grub to edit it, F10 to boot with the modified line), and see if it happens
[16:45] <iliv> setuid, it looks like so:
[16:45] <iliv> auto enp0s8
[16:45] <iliv> iface enp0s8 inet static
[16:45] <iliv> 	address 10.0.3.4
[16:45] <iliv> 	netmask 255.255.255.0
[16:45] <setuid> iliv, no gateway, no dns-nameservers?
[16:46] <iliv> no need for any
[16:46] <rattking> oh yes it did work when I set maxcpus=4.. I am wondering how I can handle that on many systems with different numbers of cpus
[16:46] <iliv> it's the secondary interface that is used in a very simplistic LAN
[16:46] <setuid> when you say "one of them doesn't work", what do you mean? It's not assigned an IP?
[16:47] <setuid> Or it's not bringing up the interface?
[16:47] <iliv> IP address is assigned to enp0s8, link is reported as up by ip a l but I can't pint this IP address or any other IP on this lan from this system until I toggle enp0s8 or restart networking via systemctl
[16:48] <setuid> You can't ping it from the same host? Or from another host?
[16:48] <iliv> setuid, note in dmesg log how enp0s3 "becomes ready" and enp0s8 never does so
[16:48] <setuid> You won't be able to ping it from another host, because you don't have the gateway set, therefore, no routes to that host, and no ARP table entries
[16:48] <iliv> I can't ping enp0s8 from another host on the same network
[16:48] <setuid> That's expected, no route, no ARP path to the host
[16:49] <setuid> But if you can't ping it from itself, that's an issue
[16:49] <setuid> back in a lunch...
[16:49] <iliv> are those all set up when I do ifdown followed by ifup?
[16:52] <iliv> in my experience it is all set up automatically as long as link is up and IP address, plus netmask, are assigned
[16:58] <iliv> setuid|lunch, see for yourself http://i.imgur.com/RuTjEFx.png This is how it looks right after the system started. Everything is there but it just doesn't work unless I toggle the interface.
[17:06] <setuid|lunch> iliv, Does it work if you run 'ifup enp0s8'?
[17:06] <iliv> setuid|lunch, if I toggle it: ifdown followed by ifup
[17:06] <iliv> setuid|lunch, if it is just ifup then no
[17:07] <ddellav> jamespage just saw your message, i will take care of those ASAP
[17:09] <iliv> setuid|lunch, here's an example http://i.imgur.com/dVC4ajl.png
[17:09] <setuid> iliv, is this a physical adapter? or virtual?
[17:10] <iliv> setuid, virtual. this system is xenial lts virtualbox guest vm. the adapter in question here is host-only type of adapter. the host is xenial lts too.
[17:11] <setuid> If you set it to 'nat', does it work?
[17:12] <iliv> I have two interfaces in this guest vm. One is enp0s3, which is the NAT interface you refer to and it always works. No problems whatsoever. The other one, which is enp0s8 we have been discussing so far, is the interface that is seemingly configured correctly but never works until it's toggled.
[17:14] <iliv> and it is host-only as I said above
[17:16] <setuid> Right, I understand that. If you change nothing at all in the guest, and just set the interface to NAT in vbox, does it work?
[17:16] <setuid> I'm trying to separate the possible issues
[17:18] <iliv> setuid, then I will end up with two NAT interfaces
[17:18] <iliv> which wil most likely isn't going to work right
[17:22] <iliv> setuid, I added this one-liner to /etc/rc.local: ifdown enp0s8 && ifup enp0s8 and it helps but it's an ugly hack obviously.
[17:25] <setuid> You can have as many NAT interfaces as you want, up to the limits of your upstream switch/dhcp server's ability to hold them in the ARP table, but that's not the issue. I'm not saying this is the solution, I'm asking to test it to see if it's an issue in the host-only networking layer (vif) vs. NAT to a physical adapter.
[17:25] <setuid> Right, that's a hack, not a solution
[17:27] <iliv> sure, give me a couple minutes to test that..
[17:31] <setuid> nacc, I'm wondering where this 'grub' issue really needs to be filed
[17:32] <jhobbs> WG 2
[17:32] <jhobbs> oops
[17:32] <setuid> nacc, Is increasing /boot/ an SOP when you need more space? Or do most people just toss the build and reinstall? Or do they nuke /boot/ partition, add /boot to / and do what I did?
[17:47] <iliv> setuid, alright so two NAT interfaces: http://i.imgur.com/LX5Q3sO.png It works.
[17:50] <rattking> setuid: when I set maxcpus=8 on a system with 8 cores I find the post-up scripts do not work again.. any ideas on how to deal with that?
[17:51] <nacc> setuid: yeah, not sure
[17:51] <nacc> setuid: i mean, in theory, increasing the size of boot maybe is normal
[17:51] <nacc> setuid: but i think most people start over when they hit it these days -- or know they don't really need a separate /boot, etc
[17:53] <setuid> nacc, I generally set /boot/ to noexec, ro, sync and make it ext2, so having it separate from / makes sense, esp. if you corrupt your rootfs, you can still boot the machine and attempt some recovery
[17:54] <nacc> setuid: sure, i can see reasons for doing it that way -- i think they are less relevant (ime) these days, but yeah, the bug might just be with usability when you try to switch /boot
[17:57] <setuid> Part misinterpretation (/ being an LVM, and me adding /boot as a directory to that partition, which grub can't boot to), and the real bug that update-grub/grub.cfg and other tools implied they were doing the right thing, but grub itself was still booting to the old -partition-, even though it was not mounted to /boot/ any longer.
[17:58] <setuid> update-grub should probably warn that it's going to create a non-bootable environment if it detects that /boot/ is on an LVM
[18:49] <iliv> setuid, what do you know! a virtio-net host-only interface works. before that I used an Intel PRO/1000 MT Desktop variety virtual NIC.
[18:49] <setuid> Sounds like a bug with the vif, file that with vbox
[18:50] <setuid> Unless you can duplicate it on vmware workstation and isolate it as an OS bug
[19:24] <JordanWP> I am using postfix on ubuntu server and my emails are relayed to MS Exchange.  Wondering how to change the From: address on the email.  Anyone?
[19:25] <JordanWP> Its currently using wordpress@%hostname%
[19:25] <bekks> Thats entirely not what a mail relay does.
[19:26] <patdk-wk> JordanWP, fix your software that creates the email
[19:26] <patdk-wk> sounds like it's wordpress
[19:26] <sarnold> there's two From headers in email, the envelope-From and then the message body From:
[19:27] <patdk-wk> no, there is only one from header
[19:27] <patdk-wk> envelope-from is not a header
[19:27] <sarnold> if your wordpress is sending with the /usr/bin/sendmail program then the -f command line option can change the envelope-from header: http://www.postfix.org/sendmail.1.html
[19:27] <sarnold> if your wordpress is setting the From: header then you need to modify the wordpress config to set the correct thing
[19:27] <patdk-wk> but it still comes down to, the software is creating the email incorrectly and needs to be fixed
[19:28] <JordanWP> okay so i need to quit looking to change anything with postfix and dig through my wordpress files.
[19:28] <sarnold> yeah, wordpress would be my starting point
[19:28] <JordanWP> okay thank you.  I will start poking around.
[21:08] <apb1963_> Aptitude is giving me an abortion when trying to install libsnmp-dev.  Details here: http://pastebin.com/AAaZgFid
[21:51] <sarnold> apb1963_: normally that means that the i386 versions don't match exactly with the already-installed amd64 versions
[21:51] <sarnold> apb1963_: try an apt-get update && apt-get -u dist-upgrade first to make sure that you've got the newest versions of everything.
[22:11] <apb1963_> sarnold: I never do dist-upgrade ...it just makes me nervous.
[22:11] <bekks> So do "apt update; apt full-upgrade" instead.
[22:11] <sarnold> apb1963_: makes sense, just do a standard upgrade instead..
[22:15] <apb1963_> well... what worries me is the fact that packages can get removed and I don't know enough to know when to be concerned. However... I'll give it a shot.
[22:15] <apb1963_> I do upgrades on a regular basis
[22:15] <sarnold> that does seem to happen a lot when people wind up installing the i386 binaries alongside the amd64 binaries
[22:16] <sarnold> for some reason some idiot software suggests something that then winds up trying to uninstall sudo.
[22:16] <apb1963_> well... that was the result of a script the asterisk related people wrote...  and yes... it's idiotic.
[22:16] <apb1963_> It looks for purged packages and then trys to install them
[22:16] <apb1963_> sacrificing the already installed version
[22:17] <apb1963_> That's where the i386 packages come from
[22:18] <apb1963_> Being the trusting sort without a good understanding of the packaging system... I ran the script.
[22:18] <apb1963_> although I'm not quite sure why they turn up in an aptitude search
[22:19] <apb1963_> oh... nvm... brain fart
[22:19] <apb1963_> strike that last sentence
[22:32] <apb1963_> how can I cat a file to pastebin directly?
[22:33] <apb1963_> I used to be able to do it.. forgot how.
[22:33] <nacc> apb1963_: pastebinit
[22:33] <apb1963_> oh yeah
[22:33] <apb1963_> :)
[22:33] <apb1963_> thx :)
[22:33] <bekks> By running: cat yourtextfile | nc termbin.com 9999
[22:33] <nacc> or that
[22:33] <apb1963_> nc... is that netcat?
[22:34] <apb1963_> output from latest script run.  It's not pretty.  http://paste.ubuntu.com/23090907/
[22:36] <sarnold> what a disaster
[22:38] <apb1963_> pretty much
[22:38] <apb1963_> afk... brb
[22:41] <apb1963_> ok
[22:41] <apb1963_> so...I guess I need to rewrite the script so that it doesn't try to install purged packages.
[22:42] <apb1963_> Perhaps use apt-get instead of aptitude?
[22:44] <apb1963_> Here's the script if yo're interested....  http://paste.ubuntu.com/23090933/
[22:44] <bekks> If your script caused that mess, I'll not use it :P
[22:45] <apb1963_> no... not for using... for fixing :)
[22:46] <apb1963_> As in... helping me to determine the correct way to do it.
[22:46] <sarnold> there's good news and bad news
[22:46] <apb1963_> lol
[22:46] <sarnold> the bad news is that script looks terrible and what it's done to your system may be insanely difficult to repair
[22:46] <apb1963_> yeah
[22:46] <sarnold> the good news is that you can very nearly do what you need form it with very little effort
[22:47] <sarnold> "apt-get install <start copy-and-psating the pacakges from the PACKAGES_DEBIAN lines>"
[22:47] <apb1963_> yeah... it's not a difficult script in general
[22:47] <apb1963_> yeah.. pretty much... I was thinking to just eliminate the search for packages since it just does stupid with the results.
[22:48] <apb1963_> And that would basically make it a plain aptitutude (or apt-get) install.
[22:50] <sarnold> these sorts of "install scripts" are always miserable compared to just preparing a package with proper dependencies, like all the other packages in the archive have done..
[22:51] <sarnold> apb1963_: here's a little script you can use to find i386 packages on your system: dpkg -l | awk '$1 ~ /ii/ && $4 ~ /i386/ {print}'   --- some may need to be there, so don't just go deleting things all over the place. but it may be helpful.
[22:51] <apb1963_> nice
[22:51] <apb1963_> thank you
[22:51] <sarnold> apb1963_: here's the ones that are already installed on my laptop http://paste.ubuntu.com/23090948/
[22:52] <sarnold> i'm surprised how many I have
[22:52] <apb1963_> http://paste.ubuntu.com/23090951/
[22:53] <sarnold> hah, that paste just has the url to your previous paste :)
[22:53] <apb1963_> hmm
[22:54] <apb1963_> o i c what happened
[22:54] <apb1963_> http://paste.ubuntu.com/23090956/
[22:55] <apb1963_> i'm running 14.04.5  you?
[22:55] <sarnold> 16.04 here
[22:57] <apb1963_> also for the record... I removed libsnmp-dev from the list because it kept aborting.
[22:58] <sarnold> I've never used aptitude enough to know how it really works but I have a feeling that this would have worked way better if the script had just blindly used apt to install everything without trying to be clever
[22:58] <sarnold> apb1963_: if you're willing to gamble a bit, try this: apt-get purge $(dpkg -l | awk '$1 ~ /ii/ && $2 ~ /:i386/ && $4 ~ /i386/ {print}')
[22:58] <apb1963_> sarnold: owwie
[22:58] <sarnold> apb1963_: that will try to delete all the i386 pacages that have :i386 in the name
[22:59] <apb1963_> sarnold: wouldn't I be better off simply letting apt-get attempt to resolve?
[23:00] <sarnold> apb1963_: maybe. your situation is .. certainly extreme.
[23:20] <apb1963_> It sure would be nice if ubuntu was distributed with essential packages pre-pinned so you couldn't hose your system unless you really knew what you were doing.
[23:21] <sarnold> yeah it'd be nice if there was some way to alllow upgrades but not uninstalls on some subset of packages..
[23:21] <sarnold> sudo has a preuninstall script that protects it
[23:21] <sarnold> but I can't imagine that anyone who has run into it also has the technical ability to make their package database happy again
[23:21] <apb1963_> and of course the asterisk guys refuse to take responsibility since it's a contributed script.  I think I"ll contribute "rm -rf /" and see if they change their policy
[23:22] <apb1963_> exactly.  The package management system is a basic nightmare.  I doubt anyone knows the proper way to use it.l
[23:29] <apb1963_> sarnold: I think what I need to do is determine which packages have both amd64 and i386 available... and then purge just those.
[23:29] <apb1963_> or at least start there.
[23:29] <sarnold> apb1963_: I think that's what the :i386 in the package name indicates
[23:30] <apb1963_> no I don't mean on my machine... I mean in the repos.
[23:30] <apb1963_> I would assume that if there is an amd64 package, that I want it.
[23:30] <apb1963_> rather than the i386 pkg
[23:30] <sarnold> that's what I mean, the :i386 bit in the package name means that that package has both amd64 and i386 variants
[23:31] <apb1963_> And if there is no amd64 available... then I get the i386
[23:31] <sarnold> heh ,are there any packages that are only i386?
[23:31] <apb1963_> huh?  so.. how does it indicate that only i386 is availalbe?
[23:31] <sarnold> now I'm curious
[23:32] <apb1963_> oh... that I don't know.  But it begs the question... why would any of my pkgs be i386 if there are amd64 versions for everything?
[23:32] <sarnold> _normally_ it would be because e.g. a game vendor only makes i386 builds of their game because they are still stuck in 2006 or something
[23:32] <sarnold> or WINE
[23:33] <apb1963_> ok... so back to your comment:  "<sarnold> that's what I mean, the :i386 bit in the package name means that that package has both amd64 and i386 variants"  How can you tell that there is no amd64 package if i386 claims both are available?
[23:34] <sarnold> apb1963_: I'm not sure, I'm trying to think of a way to figure out if there are any packages that exist for i386 but not amd64
[23:35] <apb1963_> well... wouldn't it be reported by dpkg in that script you made?
[23:36] <apb1963_> tweaked a bit of course
[23:38] <DexDeadly> Hey all, quick question.  I just setup a lamp stack.  My question is this is just local, whats a good ftp setup to do and use or any other good apache web admin
[23:38] <sarnold> that script only knows about packages that have been or are installed on the computer
[23:38] <sarnold> DexDeadly: the best ftp setup is to forget about ftp and use sftp instead :)
[23:38] <apb1963_> sarnold: yes... I'm seeing that now...  however I'm also seeing things like this:
[23:38] <apb1963_> ii  linux-libc-dev:amd64    3.13.0-93.140    amd64            Linux Kernel Headers for development
[23:38] <apb1963_> ii  linux-libc-dev:i386     3.13.0-93.140    i386             Linux Kernel Headers for development
[23:39] <apb1963_> and I think that's a pretty good indication of a problem
[23:39] <sarnold> apb1963_: there are legitimate reasons for that to happen but I don't think that script had your best interests at heart :)
[23:40] <apb1963_> no.. it really didn't.  Some clown wrote it... perhaps it worked on an earlier version.
[23:40] <sarnold> it may work fine on a system without the multiarch packages available...
[23:42] <apb1963_> Yeah... yet more confusion as to what packages should be installed.
[23:49] <apb1963_> sarnold: So I've decided to purge anomolies.  http://paste.ubuntu.com/23091091/
[23:49] <apb1963_> I did that because I saw this:
[23:49] <apb1963_> ii  gcc-4.8-base:i386                                           4.8.4-2ubuntu1~14.04.3                   i386         GCC, the GNU Compiler Collection (base package)
[23:49] <apb1963_> ii  gcc-4.9-base:amd64                                          4.9.3-0ubuntu4                           amd64        GCC, the GNU Compiler Collection (base package)
[23:49] <apb1963_> ii  gcc-4.9-base:i386                                           4.9.3-0ubuntu4                           i386         GCC, the GNU Compiler Collection (base package)
[23:51] <apb1963_> How's that look to you in terms of letting in proceed?
[23:51] <apb1963_> s/in/it
[23:52] <sarnold> apb1963_: yay, looks good
[23:52] <apb1963_> ty
[23:52] <sarnold> apb1963_: thogh there's that 'skype'. If you want it you may want to reinstall skype again once you're done..
[23:53] <apb1963_> Yeah... would be nice if I could say  yes to all except skype
[23:53] <apb1963_> or  yes this no that.
[23:58] <apb1963_> I was about to remove the i386 version but since it had so many pkgs unrelated to asterisk (like adobe reader) I decided to see what the amd64 version looked like. http://paste.ubuntu.com/23091113/
[23:58] <apb1963_> sarnold: ^^
[23:59] <sarnold> apb1963_: what does dpkg -l libc6 show?
[23:59] <apb1963_> ii  libc6:amd64                         2.19-0ubuntu6.9        amd64                  Embedded GNU C Library: Shared libraries
[23:59] <apb1963_> ii  libc6:i386                          2.19-0ubuntu6.9        i386                   Embedded GNU C Library: Shared libraries