=== MatthewsFace is now known as MatthewsFace[SEA [01:36] Greetings [01:36] new here [01:36] can anyone tell me if vnc works on latest ubuntu server [01:37] It loads at startup but I cant seem to connect to it [01:38] I am trying to run a headless server for mythtv [02:01] http://www.ubuntu.com/download/cloud/install-ubuntu-openstack [02:01] does this need to be on every node? === zz_DenBeiren is now known as DenBeiren [02:39] harushimo: no; maas and landscape will know how to install ubuntu and openstack on the other nodes [02:40] those instrucations are reflecting for the main node [02:40] I mean instructinos [02:40] I mean instructions [02:40] harushimo: yeah; thanks for the link, that looks really useful :) [02:40] not a problem [02:41] This is my 6th attempt at openstack [02:41] I get closer everytime [02:41] I know the feeling. [02:42] one tip about maas.ip [02:43] to get the page: you need the ip of the VM and load it in the browser [02:44] whenever I lose one of my machines, I always use nmap to find it again; nmap 192.168.1.0/24 -p 80 or something like that, to scan the subnet for port 80 [02:44] interesting [02:45] (I almost always scan for ssh on 22 instead of httpd on 80, since the machines I care about more often have ssh open, but 80 might be faster to find a machine with web control thingy live..) [02:45] right === beisner- is now known as beisner [03:47] Need some help configuring a headless server [03:47] cant seem to get vnc to load properly at boot [03:51] boukmandutty: what errors are you gettting? [03:52] I get this [03:52] Active: active (running) since Mon 2015-05-04 22:40:59 CDT; 10min ago [03:52] Process: 933 ExecReload=/bin/kill -HUP $MAINPID (code=exited, status=0/SUCCESS [03:52] oops [03:52] typo [03:52] vncserver.service - (null) [03:52] Loaded: loaded (/etc/init.d/vncserver) [03:52] Active: active (exited) since Mon 2015-05-04 22:41:06 CDT; 11min ago [03:53] I notice it exits when booting up [03:53] I had set it to load at boot [03:53] boukmandutty: anything in the logs? [03:54] But I am then able to load it once I log into lxde [03:55] 04/05/15 22:41:05 Xvnc version TightVNC-1.3.10 [03:55] 04/05/15 22:41:05 Copyright (C) 2000-2009 TightVNC Group [03:55] 04/05/15 22:41:05 Copyright (C) 1999 AT&T Laboratories Cambridge [03:55] 04/05/15 22:41:05 All Rights Reserved. [03:55] 04/05/15 22:41:05 See http://www.tightvnc.com/ for information on TightVNC [03:55] 04/05/15 22:41:05 Desktop name 'X' (ubunserver:1) [03:55] 04/05/15 22:41:05 Protocol versions supported: 3.3, 3.7, 3.8, 3.7t, 3.8t [03:56] 04/05/15 22:41:05 Listening for VNC connections on TCP port 5901 [03:56] Font directory '/usr/share/fonts/X11/Type1/' not found - ignoring [03:56] Font directory '/usr/share/fonts/X11/75dpi/' not found - ignoring [03:56] Font directory '/usr/share/fonts/X11/100dpi/' not found - ignoring [03:56] ** Message: main.vala:99: Session is Lubuntu [03:56] ** Message: main.vala:100: DE is (null) [03:56] ** Message: main.vala:110: No desktop environnement set, fallback to LXDE [03:56] Xlib: extension "RANDR" missing on display ":1". [03:56] ** Message: main.vala:131: log directory: /home/boukmandutty/.cache/lxsession/Lubuntu [03:56] ** Message: main.vala:132: log path: /home/boukmandutty/.cache/lxsession/Lubuntu/run.log [03:56] I am running ubuntu server 15.04 [03:58] The full message when I run "systemctl status vncserver.service -l" is as follows [03:58] .. please use a pastebin :) [04:02] http://pastebin.com/nfAdbkyC [04:03] thanks :) [04:03] Sorry about the flooding above [04:04] Havent used irc in a bit [04:04] what's in /home/boukmandutty/.vnc/ubunserver:1.log [04:07] http://pastebin.com/QV66TZFK [04:09] boukmandutty: it looks like it shuold be working fine; it starts an X server on :1 listens for connections on tcp port 5901, starts the lxde window manager / environment.. [04:09] boukmandutty: does netstat -anp | grep 5901 show it listening? [04:12] Xtightvnc as listening but I am wondering if that is because I restarted it after loggin into lxde [04:13] ah could be [04:13] i am going to reboot and try again [04:18] Interesting it still shows cp 0 0 127.0.0.1:5901 0.0.0.0:* LISTEN 1048/Xtightvnc [04:18] but I cant connect from another machine [04:21] hmm [04:21] it's bound to localhost; I wonder if a public IP is available when it is starting? [04:22] Do you think I should try setting a static ip instead of DHCP ? [04:22] boukmandutty: check the configs, see if it is set to bind to localhost or if it is set to bind to * or 0.0.0.0 -- this one may be a bug... [04:22] boukmandutty: not yet [04:22] boukmandutty: if you want to for other reasons, go ahead, but this might be easier to reproduce with dhcp.. [04:23] ok [04:25] You mean the configs for vncserver? [04:26] yeah [04:28] Neither .vnc/xstartup nor /etc/init.d/vncserver have say anything about an ip address [04:29] is there an /etc/defaults/*vnc* or /etc/*vnc* that would have such an address? [04:31] I dont seem to have either of those folders [04:32] trying to search to see where ubuntu server keeps the equivalent [04:35] boukmandutty: time for me to bail, have fun, good luck :) [04:35] hey [04:35] thanks for the help man [04:35] appreciate it [04:35] you're welcome, I just wish we'd cracked this one first. [04:35] know of any other way to adminsiter headless server? [04:36] any other program i can use? [04:36] I just ssh in [04:36] ok will try that [04:36] ssh X forwarding suffices for the handful of times I might want a GUI running on another machine.. [04:36] ssh -X -Y hostname virt-manager & for example [04:36] i will try ssh in and then running the vnc for now [04:37] since it loads after startup [05:27] I just created an amazon ec2 which requires me to login to ssh via a key.pem. that I am able to but now I want to add new users but not share the key. What I did was, created a new user and gave him password. Tried to login by ssh again but now it says "Permission denied (publickey)." why so? [06:48] so I rm ed a log file, mail.log and it doesn't seem to have come back yet [06:49] I'm trying to troubleshoot an issue === Lcawte|Away is now known as Lcawte === Lcawte is now known as Lcawte|Away [07:46] I wonder WHY everyone can read a file that I created......-rw-rw-r-- 1 developer developer-group 5 May 5 06:36 abc . I just used nano. if this is default behaviour of linux to give world wide read persmissions to any file that is creatd by any user; its bad. no? [07:48] Voyage: its configurable - read up about umask. umask 006 will make it not world readable. [07:48] no, it is good in collaborative multiuser environments ... you can always create a folder that only you can read and put your secret stuff in it as a user ... or as a sysadmin you can change the systemwide umask [07:49] hm. but I though everything in /home/ is private and should not be implicitily r by all [07:50] what makes you think that? === bradm_ is now known as bradm [07:50] see /etc/login.defs there is an explanation in the comment section above the UMASK option [07:50] I created a file, it has -rw-rw-r-- <-- this makes me thingk [07:51] that makes you think its private? [07:51] the /home/user. is always private.. thats what home means [07:52] no, it really doesn't. [07:52] hm [07:52] it only means it is your space on the filesystem, nothing more [07:52] you're making an invalid assumption, things are default setup as world readable - if you want to change it, fix the umask - either in /etc/login.defs, or in your private login files or similar [07:52] if it is private or not is up to the sysadmin to decide === liam_ is now known as Guest74585 [08:35] Good morning. [08:54] morning [09:01] Hi, is there a way with upstart to force a job to run before runlevel jobs kick in ? [09:02] I need to make sure that kdump runs before the Runlevel 2 jobs start, otherwise CEPH OSD start during a kernel dump [09:55] hey... anybody know where the lxc container config files are when lxd is used? === Lingo_ is now known as IronDev === Lcawte|Away is now known as Lcawte [12:02] jamespage: I'm not always online but have you seen the bug report for the duplicate manpage in the cloud archive repository for ceph and ceph-deploy? [12:41] hmm, i'm still having throughput issues.. ahhhh lol [12:44] Hi.. Where should I put my custom init script in ubuntu 15.04? /etc/systemd/system/my-custom.service or ..? [12:45] ozanhazer: I believe that's the right place, yes. I'm not sure though. [12:49] hmm /etc/systemd/system/multi-user.target.wants is better maybe.. thanks rbasak [12:53] I'm copying from one NVMe drive to another NVMe drive over NFS @ 10Gbps and I'm only getting 426MB/sec [12:54] boo [12:54] Teduardo: To me that is still quite impressive ;) [12:54] unless you were the one that paid $2k each for the NVMe drives [12:56] should easily be able to saturate 10G [12:56] heh? [12:56] why? [12:56] you have nvme speed to worry about, network latency, and tcp [12:57] did you tune your tcp for 10gbe? [12:58] No, I assumed that 14.04 was pretty much tuned [12:58] Teduardo: it is not [12:59] Teduardo: What 10Gnics do you have? [12:59] still iperf shows sustained line rate [12:59] tuning for 10gbe would cause 1gbe to suffer [12:59] X710DA-2 [12:59] Teduardo: For filecopy I would use dd and nc [12:59] if its single, very large files [12:59] im still just testing/benchmarking everything [13:01] Teduardo: And your subsystem can handle suntained writes/reads at 1GByte/sec? [13:01] One of the servers is dual E5 2620 and the other is a dual E5 2620v3 [13:02] i would imagine that the PCIe in those can do 1GB [13:03] But you said you copied files? [13:04] over nfs [13:04] yes. [13:04] oh, nfs [13:04] Where does those files end up? [13:04] mounted nvme => nfs => mounted nvme [13:04] that changes things a good amount [13:04] nfs isnt exactly superfast, and it is the nfs-kernel-server? [13:05] but even doing dd bs=8m if=/run/shm/testfile of=/nvme/testfile it only does 500MB/sec [13:05] which is insane [13:05] from ram to an nvme [13:05] 500MB? [13:05] (locally I mean) [13:05] on one of the systems [13:06] doesn't sound like your using nvme [13:06] Teduardo: you are talking about the PCIe SSD from intel? Newly released? [13:06] yes, the DC P3700 [13:06] what slot in your motherboard is it in? [13:06] and what motherboard? [13:06] ok, 400GB? [13:06] yeah, its the 400GB cause those were the only ones I could get. [13:07] That has 900MB/s sekvential write theoretical max [13:07] on subsequent runs writing from local ram to local NVM sped up to about 685MB/s [13:08] okay, reading from the NVM to the ram went at 1.3GB/s [13:08] Where they will shine is for random writes and reads at the time time [13:08] yes [13:08] 2,2GB/sec theoretical max [13:08] okay so at least i know the NVM is reading at least 10Gbps [13:08] yes [13:09] now i just need that to go through the NIC to the other machine. [13:09] How are they connected? p2p or via switch? [13:09] just a da2 cable [13:09] ok [13:11] seems like from nvme to the raid-10 fs locally it does about 480MB/s which i doubt is a real number [13:12] Why not? [13:12] 8x4TB drives with default stripes... [13:12] Raid-10 on mechanical drives, have very fast sequential write/read performance. [13:12] i thought it would be like ~300 [13:13] or i would be happy with a sustained 300 [13:14] the actual issue I am having is sustained throughput between a 8x4TB R-10 on machine a and a 8x4TB R-10 on machine B. the throughput in rsync stops at 200MB/s [13:14] and drops down to 80MB regularly [13:15] Teduardo: What switches are you using for rsync? [13:15] And how much data and how many files? [13:15] just rsync -a --progress src_file dst_file [13:15] ok [13:16] but like i said it's over nfs [13:16] rsync is not so good for transfers of very large files 1Gbyte+ - since its hashing will use a lot of resources then [13:16] so try to slap on --whole-file [13:16] and see if its more liniar [13:16] 8 4tb disks should be approx 4*120MB/sec [13:16] sounds like 480MB is valid [13:17] rsync depends on filesize [13:17] the smaller the files, the slower it will go [13:17] OpenTokix: it started at 218 then dipped down to 80 and now it's just bouncing between 218 and 80 every few seconds. it's a single 50GB file for this test. [13:18] Teduardo: ok, try --whole-file - but rsync is not the best tool for that [13:27] 174.32MB/s [13:28] iirc rsync doesn't rsync on a local system. it becomes copy. [13:30] if you are copying 1 file, using cp is going to use much less resources and be faster. [13:31] yea, this is just a test to see what the hardware is capable of [13:32] so that i can then yell at the software provider to see why it runs at 30MB/sec [13:32] from the man page for the -W option: "This is the default when both the source and destination are specified as local paths" [13:32] bonnie++ is a nice at telling of what hardware is capable [13:33] well, its a client/server setup [13:33] ok. sounded like src_file and dst_file were both local paths. rsync doesn't know that one of those is NFS and do any magic. [13:33] ah [13:35] jamespage, ping [13:35] rbasak: ok, i'll take a look at the code (next time it bugs me, which should be soon), thanks [13:35] hey jcastro [13:35] http://summit.ubuntu.com/uos-1505/2015-05-05/ [13:35] there are two openstack sessions today [13:36] (they are currently white on the schedule, I am trying to make them green) [13:37] jcastro, hmm - not sure these are really uds topics? [13:37] well they were submitted and pat had me approve them [13:37] not sure what to do now [13:37] jcastro, who by? [13:38] https://launchpad.net/~su-zhang submitted the congress one [13:38] https://blueprints.launchpad.net/rally/+spec/rally-neutron-metering-scenarios is the other one [13:46] jcastro, I think these are confused - they should be targetting the openstatck design summit [13:46] not the ubuntu developer summit [13:46] rally is an upstream openstack project [13:46] and the congress/keystone conversation is similar to that - upstream oriented [13:47] william_home, hello [13:47] jamespage: hello [13:48] william_home, this one - https://bugs.launchpad.net/cloud-archive/+bug/1450175 ? [13:48] jamespage, should I cancel them then? [13:48] and mail the individual folks? [13:48] jcastro, I would yes [13:48] ok on it [13:48] jamespage: yes [13:48] redirect them to the openstack design summit submissions process [13:49] ack, I'll sort it [13:49] jamespage, are there any sessions you plan to have? [13:49] i wish there was a way to know what resource it's running out of when the throughput backs off [13:50] jcastro, we have two on thursday [13:50] for openstack [13:51] * jcastro nods, ok [13:53] jcastro, just checking with the ceph guys to see if they want to have one [13:59] william_home, I guess we could include an updated ceph-deploy - I am making the assumption that the newer version does not exhibit the same problem :-) [14:01] jamespage: I have created from github a new package https://github.com/ceph/ceph-deploy [14:02] but I guess that it woul be better that ceph-deploy gets its man page in its own package rather then that it is deployed in here: https://github.com/ceph/ceph/tree/master/doc/man/8 [14:03] oh and the package from github does not contain any man page :) [15:17] rbasak, or anyone, i'm looking at getting python2 out of cloud-image [15:17] i know that landscape-client and landscape-common are one of the reasons. [15:17] i apt-get remove them. [15:18] then, [15:18] apt-get remove libpython2.7 python [15:18] http://paste.ubuntu.com/10990424/ [15:18] how can i know why each of those things is there. [15:18] as for example, i can: [15:19] apt-get remove python-twisted-web [15:19] and there are no additional things removed.. ie, it doens't look like its depended on. [15:19] but 'apt-get autoremove' doesn't clean anything [15:19] i know that vim is holding some stuff, but owuld like to know what else... [15:20] looking for a way to do such a thing. [15:21] smoser: http://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.vivid/rdepends/python2.7/libpython2.7 maybe? [15:26] but how do i know why 'python-serial' was there in the first place [15:28] smoser: http://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.vivid/cloud-image ? [15:28] python-serial isn't in that pastebin [15:28] http://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.vivid/rdepends/twisted/python-twisted-core though shows how that got in. [15:29] If you want libpython2.7 out of main, then http://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.vivid/rdepends/python2.7/libpython2.7 shows you what you need to clear. [15:29] (I think it follows both a and b for "a | b" though, so there will be some extra stuff in there) [15:30] If you only care about particular seeds, then just search the file for instances of the seed you're after. [15:31] 'apt-get autoremove' won't work because IIRC all the default stuff gets marked as "manually installed". [15:32] http://people.canonical.com/~ubuntu-archive/germinate-output/ubuntu.vivid/rdepends/twisted/python-twisted-core suggests that python-twisted-web got installed because of landscape-client. [15:33] But since autoremove doesn't work, I think a better approach is to follow the tree from germinate-output rather than trying to follow it with apt on a live system. [15:35] You could potentially write a 'for installed package { if package not in seed { mark package as automatically installed } }' sort of a script, but I don't know if that actually gains you anything. [16:07] is there a way to see what packages would be upgraded and to what version with a do-release-upgrade without any changes made to the system? [16:21] Onionnion, i dont know if theere is a dry-run for it or not. but you could: [16:21] back up /etc/apt/sources.list /etc/apt/sources.list.d [16:22] sed 's,,' -i /etc/apt/sources.list /etc/apt/sources.list.d/* [16:22] apt-get update [16:22] apt-get dist-upgrade --dry-run [16:22] replace backd up sources.list [16:25] so to just adjust the sources file to point to the newer release and dry-run in apt-get? [16:33] smoser: do-release-upgrade -s ? [16:33] that'd ake sense. [16:33] i didn't know if that worked any more [16:33] but that'd do a ton of download and such if it id id [16:52] I hate how vmware workstation's virtual network nat service will just randomly shit on you. [17:14] hey is jeos and ubuntu-vm-builder dead? is there a tool that supersedes that combo? [17:15] mbroadst: cloud images, uvtool and mount-image-callback. [17:15] mbroadst: or alternatively the ubuntu core images I suppose [17:15] and snappy. [17:16] yeah, snappy ubuntu-core [17:17] I got the impression snappy was sort of a coreos + docker competitor? [17:19] if i have a virtualbox instance of ubuntu server 14.04, can i access the default webcam of the server? [17:20] Depends on how it is connected to your computer, technically. [17:20] rbasak: I guess what I'm looking for really is something that allows me to automate the creation of a raw ubuntu-baed disk image [17:21] ubuntu-server based image that is [17:21] mbroadst, snappy is embedded, cloud ... soon the base for desktop and phone too ... and yes, you can use a docker framework on top of it === liam_ is now known as Guest95660 [17:21] essentially it is the future design of ubuntu [17:22] mbroadst: start with a cloud image. Use mount-image-callback to modify it. Create /var/lib/cloud/seed/nocloud/ inside it, and userdata and metadata files for cloud-init. [17:22] mbroadst: then just boot that image. [17:22] mbroadst: see the output of "cloud-localds" for an example of userdata and metadata files [17:23] rbasak: so you're suggesting cloud images + uvtool over snappy then [17:23] mbroadst: depends on what you're trying to do. [17:23] snappy is a great new model with many improvements over the traditional model. [17:23] snappy currently still means you will have to roll your own project snap packages [17:23] But it involves having maintained snaps for everything you need (or you maintain them yourself) [17:23] Right. [17:24] If you already have stuff that fits the traditional model, you can use cloud images instead. [17:24] if you are willing to invest into that, snappy should be your choice, else use cloud images [17:24] well what exactly constitutes a snap in this context, a single app like mysql, or a segment of the file system (/opt/something) [17:24] mbroadst: if you just want to fire up a VM, uvtool wraps that all for you. [17:25] a bundle ... [17:25] A segment of the file system [17:25] I suppose you could bundle everything you want into one snap. [17:25] (that you maintain yourself) [17:25] you can see a snap like a bundle of services confiugured in your preferred way [17:25] and each of the snap are like a ro aufs overlay or something [17:25] But, for example, I don't want to maintain my own mysql snap - I want to use existing mysql packaging [17:27] i.e. i could roll lamp_ogra.snap which has my personal setup of mysql, apache etc in it ... and always re-use it ... installing it will run it in a confined corner of the filesystem ... if i messed up security or whatever nothing can break out of it ... if i add a fix and the fix is broken, snappy allows me seamless rollback of the whole snap [17:28] ogra_: anything on the roadmap for Juju integration? [17:28] I want separate snaps with relations between then :) [17:28] them [17:28] i guess so, i dont know the snappy roadmap yet (just changed teams) [17:35] so, if i have ubuntu 14.04 server in virtualbox, can i access the webcam? [17:35] ;'( [17:35] dunno, ask virtualbox [17:36] For a USB webcam? USB passthrough is not uncommon. KVM can do it I think. Maybe virtualbox can too? [17:36] ogra_: seems like snappy requires a decent amount of user interaction. I'm trying to automate the build of reproducible images. Ideally I'd be using something like bitbake/yocto, but ubuntu certification is required by the customer [17:37] well, snappy surely comes closer to a bitbake or yocto binary result here [17:37] okay maybe I'm missing the automated integration of snaps [17:38] I'm checking out the webcam example [17:38] but it definitely is at a point where you have to do a bit more work still [17:38] it seems like I can automate the building of the snaps [17:38] but actually loading them in is manual [17:38] I can't just pop out a raw disk image with the snaps in them right? [17:38] I presume the ability to build a snappy+snaps image is a primary use case for IoT. I would assume that you can do that. [17:39] mbroadst: so maybe try #snappy? They know far more about this than just me and ogra_ :) [17:39] ah, right :) [17:39] sorry got a little lost in the twenty tabs of concurrent reading over here :) [17:39] I am curious about the answer though [17:44] has anyone a preseed template for softraid ? === soren_ is now known as soren [19:24] is it me, or is this documentation simply wrong? http://upstart.ubuntu.com/cookbook/#pre-start [20:05] jrwren: what part/ [20:06] it does not seem to mix with dpkg [20:06] i have a debian/upstart script and the stop; exit 0; causes the invoke-rc.d to return a non-zero exit which dpkg hates [20:07] I'd thought this was a common pattern, but it seems it isn't. [20:09] ahh :) [20:10] any suggestions or places to look? [20:11] no, sorry === DenBeiren is now known as zz_DenBeiren [20:39] decided to override the dh_installinst generated postinst. it works. === ebonics_ is now known as ebonics === Lcawte is now known as Lcawte|Away === DenBeireo is now known as DenBeiren === neunon_ is now known as neunon === Lingo is now known as IronDev