2014/03/06/#cloud-init.txt

[00:42] <smoser> harlowja, well, i'd replace any usage of the templates that we have in trunk to use the new format.
[00:43] <harlowja> smoser right, i adjusted all the ones in /templates
[00:43] <smoser> and those templates would in some way declaratively state that they were of this new type.
[00:43] <harlowja> yup
[00:44] <harlowja> check out the merge smoser when u get some time
[00:44] <harlowja> *review/merge
[00:46] <smoser> harlowja, thanks for all your help.
[00:46] <harlowja> np boss 
[00:46] <harlowja> :)
[00:46] <harlowja> like 100 distros in cloud-init now
[00:46] <harlowja> haha
[00:46] <harlowja> pretty interesting
[00:47] <smoser> horah!
[00:47] <harlowja> def
=== ctracey is now known as ctracey|away
[04:09] <oobx_> howdy? I'm stuck with vcloud director and am trying to add cloud-init to my vms because vcloud's vm customization is picky about the OS versions that it supports.
[04:10] <oobx_> I've created a user-data.txt file and genisoimage'd it, uploaded, attached, and booted.  But, /var/log/cloud-init.log doesn't show it noticing the ISO
[04:11] <oobx_> i'm using the EPEL 6.3 version of cloudinit on centos 6.5
[04:21] <oobx_> any tips?  It doesn't show "Alt cloud" as a datasource in the log, just nocloud, configdrive, ec2 and OVF.  I guess I need to find the 7.5 rpm or build one, right?
=== harlowja is now known as harlowja_away
[04:51] <oobx_> i just noticed that OVF requires userdata to be in XML. I just used the make-iso script to convert my user-data into bases64 and shove it into the XML file.  I'll try it out tomorrow, since I'm unable to upload the ISO right now.
=== harlowja_away is now known as harlowja
=== harlowja is now known as harlowja_away
[09:43] <harmw> seanwbruno: you've had a chance to test the freebsd stuff?
[09:45] <harmw> smoser: anything needed from my part to have the pending merge request in cirros be fullfilled?
[09:45] <harmw> *accepted
=== oobx_ is now known as oobx
[14:00] <smoser> harmw, i'll try to do that today.
[14:00] <smoser> i'm really sorry. i want to run it in lxc once. (even though i made you boot an unholy ubuntu system to try that :)
[14:31] <harmw> oh it worked on my fedora system :) 
[14:31] <harmw> and sure, np
[17:15] <harlowja> any RH folks around? i was wondering what version of cloud-init will be in rhel7
[17:15] <harlowja> *and more importantly with what cloud.cfg
[17:16]  * harlowja might save me some time rebuilding a rhel7 rpm if i can avoid building it in the first place :-P
[17:37] <harmw> harlowja: does it even ship with ci? I dont see it at either ftp://ftp.redhat.com/redhat/linux/enterprise/6Server/en/os/SRPMS/ or the path for the 7beta
[17:37] <harlowja> hmmm, was talking to the folks here, and i think they said it does, let me double check, i just got access to a rhel7beta
[17:37] <harmw> I could be wrong ofc
[17:38] <harlowja> k, think u are right, doesn't seem in the rhel-beta repo, wonder where the guy saw it
[17:38] <harmw> perhaps epel, since it's include there
[17:40] <harmw> http://mirror.nl.leaseweb.net/epel/6Server/SRPMS/cloud-init-0.7.4-2.el6.src.rpm
[17:40] <harmw> thats for rhel6/centos6
[17:41] <harmw> http://mirror.nl.leaseweb.net/epel/beta/7/SRPMS/cloud-init-0.7.2-8.el7.src.rpm
[17:41] <harlowja> k, found it
[17:41] <harlowja> http://dl.fedoraproject.org/pub/epel/beta/7/SRPMS/repoview/cloud-init.html
[17:41] <harmw> and thats 7
[17:41] <harmw> ah yes harlowja
[17:41] <harlowja> ya, looking inside it, seeing what they did :-P
[17:41] <harmw> hhe
[17:42] <harlowja> ya, it seems like they just took over the fedora one :-P
[17:42] <harlowja> distro: fedora...
[17:42] <harlowja> in cloud.cfg
[17:43] <harmw> ah yes, I believe I've patched that in my private repo to read rhel or something :p
[17:43] <harlowja> ya, some systemd changes also
[17:43] <harmw> are those patches fedora specific or should thy be included 'upstream'?
[17:43] <harlowja> not 100% sure :-P
[17:43] <harlowja> they don't seem so specific
[17:44] <harmw> the nodevconsole patch looks interesting
[17:44] <harlowja> ya
[17:46] <harlowja> hey, at least rhel7 is 'Python 2.7.5'
[17:46] <harlowja> lol
[17:46] <harlowja> :-/
[17:47] <oobx> harlowja: I was tinkering with my desktop and may have missed help directed at me.  Thanks for the cloud-init epel links for CENTOS 7.  Now, I'm seeing 7.4 instead of 6.3 in http://dl.fedoraproject.org/pub/epel/6/x86_64/
[17:48] <harlowja> :)
[17:48] <harlowja> there u go
[17:48] <harmw> smoser: got that lxc container up already? :p
[17:58] <smoser> harmw, so in my test here.. i see the disk output of the host.
[17:58] <harmw> you do? hm
[17:58] <harmw> proc/partitions has indeed all the hosts stuff
[17:59] <harmw> but lsblk didn't
[18:09] <smoser> i'll lok at this a bit more. 
[18:09] <smoser> i think i might use lscpu
[18:09] <smoser> as it can tele me cores x sockets x threads
[18:10] <harmw> and thats included in cirros already?
[18:10] <smoser> yeah.
[18:10] <harmw> damn
[18:10] <smoser> and the lslk might differ between yours and mine.
[18:10] <smoser> (the lxc container might differ)
[18:10] <harmw> yea could be the case
[18:10] <smoser> but at this point whatever we see is probably not valid.
[18:27] <smoser> harmw, lscpu --parse
[18:27] <smoser> that looks like it would be nice. 
[18:27] <smoser> but it seems to not work :)
[18:27] <smoser> oh. i see.
[18:28] <smoser> it does. but doesn't give as much info as you'd like.
[18:28] <harmw> damn, nifty output
[18:29] <harmw> though it's all zeroes whn running --parse
[18:32] <smoser> well, its not. 
[18:32] <smoser> thats only for the first cpu
[18:32] <smoser> and you only have 1
[18:32] <smoser> :)
[18:32] <harmw> indeed I have
[18:32] <smoser> i didn' realize cirros /bin/sh supported ${FOO//from/to}
[18:32] <smoser> that is not posix
[18:34] <smoser> awk '$1 == "MemTotal:" {print $2}' /proc/meminfo
[18:34] <smoser> you can do the divide in awk too if you wanted. and even then ditch the echo.
[18:55] <smoser> harmw, http://paste.ubuntu.com/7045706/
[18:55] <smoser> that prints out:
[18:55] <smoser> arch: x86_64
[18:55] <smoser> total: 16 cpu @ 1600.000 MHz
[18:55] <smoser> cores/sockets/threads: 4x2x2
[19:12] <harmw> arch: x86_64
[19:12] <harmw> total: 1 cpu @ 2009.101 MHz
[19:12] <harmw> cores/sockets/threads: 1x1x1
[19:12] <harmw> nice
[20:10] <harmw> === system information ===
[20:10] <harmw> Arch: x86_64
[20:10] <harmw> CPU(s): 1 @ 2009.101 MHz
[20:10] <harmw> Cores/Sockets/Threads: 1/1/1
[20:10] <harmw> Virt-type: AMD-V
[20:10] <harmw> Hypervisor:
[20:10] <harmw> I think I like that output a little better
[20:11] <harmw> (and to figure out the hypervisor I'd probably only have dmidecode as option)
[20:14] <smoser> harm. tha tlooks fine to me.
[20:14] <smoser> one cleanup to what i had, you should initialize variables to 'NA' or something.
[20:17] <smoser> harmw, http://paste.ubuntu.com/7046165/
[21:09] <harmw> smoser: pushed
[21:09] <harmw> and I realy can't wait for dmidecode... :)
[21:10] <smoser> harmw, dmidecode is "interesting"
[21:10] <smoser> i recently fixed two bugs in cloud-init to *not* use dmidecode
[21:10] <harmw> haha
[21:10] <smoser> as on arm, it crashes vms
[21:10] <smoser> and even, reportedly, crashes hardware.
[21:10] <harmw> ouch
[21:10] <smoser> ie, you run that command and "poof" your hardware goes MIA.
[21:11] <harmw> well, perhaps /dev/mem is just to delicate :p
[21:11] <harmw> anyway, time for a new cirros release?
[21:13] <harmw> hm
[21:13] <harmw> dmesg|grep DMI:
[21:13] <harmw> looks like a simple way of telling which platform we're on
[21:14] <harmw> [    0.000000] DMI: Red Hat Inc. OpenStack Nova, BIOS Bochs 01/01/2011
[21:14] <harmw> kindof
[21:14] <harmw> dmesg|grep DMI:
[21:14] <harmw> DMI: Supermicro X7SPA-HF/X7SPA-HF, BIOS 1.0b    01/19/2010
[21:16] <harmw> (or var/log/messages for that matter)
[21:17] <harmw> smoser: I think I like that line as part of the extended info stuff
[21:17] <harmw> thoughts?
[21:18] <harmw> cut until BIOS shows up that is
=== ctracey|away is now known as ctracey
[21:30] <smoser> $ dmesg | sed -n 's/.*[ ]\+DMI:[ ]\+//p'
[21:30] <smoser> LENOVO 7417CTO/7417CTO, BIOS 7UET91WW (3.21 ) 12/06/2010
[21:31] <smoser> harmw, lets do something simple now. and get you some of the dat ayou want.
[21:31] <smoser> and then work on getting something nicer.
[21:43] <harmw> ok 
[21:43] <harmw> grep DMI /var/log/messages | sed 's/.\+] DMI: \(.\+\), BIOS.\+/\1/'
[21:47] <smoser> no need for grep. 
[21:47] <smoser> sed is doing that above.
[21:47] <smoser> and in yours too.
[21:48] <smoser> $ sed -n 's/.*[ ]\+DMI:[ ]\+//p' /var/log/kern.log
[21:48] <smoser> LENOVO 7417CTO/7417CTO, BIOS 7UET91WW (3.21 ) 12/06/2010
[21:49] <smoser> use '-n'
[21:51] <harmw> well, using grep at first seems a little faster 
[21:54] <harmw> $ grep DMI: /var/log/messages | sed -n 's/.*[ ]\+DMI:[ ]\+\(.\+\), BIOS.*/\1/p'
[21:54] <harmw> Red Hat Inc. OpenStack Nova
[21:57] <smoser> really?
[21:57] <smoser> are you sure?
[21:59] <smoser> yeah. wow. it is. 
[22:00] <harmw> :)
[22:00] <harmw> I'm pushing it 
[22:05] <harmw> http://paste.ubuntu.com/7046723/ thats how it looks like
[22:08] <harmw> idealy, the sudo fix for reading ssh keys shouldn't have gone into this branch
[22:13] <harmw> ok, and using awk to get the ram size
[22:15] <smoser> right. you dont need sudo there.
[22:15] <smoser> right?
[22:15] <harmw> well, not at boot time no
[22:15] <smoser> ah. i see.
[22:15] <harmw> but you do if you run the cirros-status manually
[22:15] <harmw> for whatever reason
[22:15] <smoser> right. other option is just to output only what is avilable as non-root.
[22:16] <smoser> and say "if you were root, you'd know more"
[22:16] <harmw> lets merge this, release cirros and then change that :p
[22:17] <smoser> :)
[22:18] <smoser> i have to run.
[22:18] <harmw> ok :)