15:00:07 <primeministerp> #startmeeting
15:00:08 <openstack> Meeting started Tue Jul 31 15:00:07 2012 UTC.  The chair is primeministerp. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:09 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:21 <primeministerp> #topic Hyper-V
15:00:34 <primeministerp> Howdy everyone
15:00:56 <primeministerp> I sent out the topics for discussion to the list
15:01:16 <primeministerp> •	Status of the Mock Hyper-V Compute
15:01:16 <primeministerp> •	Move  from Essex to Folsom
15:01:16 <primeministerp> o	Installer is for essex integration
15:01:16 <primeministerp> •	Hyper-V Testing Procedures
15:01:16 <primeministerp> o	Dev stack for hyper-v testing
15:01:19 <primeministerp> o	Chef for Windows automation
15:01:39 <primeministerp> We'll go through those now
15:02:01 <primeministerp> #topic Hyper-V Moc driver
15:02:24 <primeministerp> Alessandro and I have been working to put together a plan of attack
15:02:37 <primeministerp> alexpilotti: care to add some input
15:03:22 <primeministerp> In a nutshell he is in the process of the implementing the wmi moc interface
15:05:09 <primeministerp> alexpilotti: needs a minute
15:05:16 <primeministerp> we'll come back to this
15:05:53 <primeministerp> #topic Upstream Code integration
15:06:25 <primeministerp> Work needs to begin to integrate the upstream code
15:06:38 <primeministerp> pnavarro: is this something you can begin looking at
15:07:16 <primeministerp> pnavarro: we know we have issues w/ the glance, python-novaclient
15:07:24 * primeministerp looks up the other
15:07:43 <pnavarro> primeministerp: yes we have, some packages are not compatible with windows
15:07:52 <primeministerp> pnavarro: right
15:07:58 <primeministerp> so wee need to flush those out
15:07:59 <alexpilotti> hi guys!
15:08:04 <primeministerp> and have the code put in to exclude it
15:08:20 <primeministerp> alexpilotti: great
15:08:46 <alexpilotti> pnavarro: hi!
15:08:54 <pnavarro> hi alexpilotti
15:08:54 <primeministerp> pnavarro: i'll be able to flush this out once testing is underway
15:09:07 <primeministerp> pnavarro: and hopefully supply useful bits for bugs
15:09:37 <primeministerp> pnavarro: can you expand on what you know?
15:10:45 <primeministerp> so
15:10:52 <primeministerp> no hitesh today
15:10:55 <primeministerp> ok
15:11:08 <primeministerp> so I guess what needs to happen is people need to manually install the bits
15:11:26 <primeministerp> to see exactly these incompatibilites exist
15:11:31 <alexpilotti> pnavarro: do you have a list of the packages that were incompatible in your setup?
15:11:35 <primeministerp> and then notify the people upstream
15:11:38 <primeministerp> alexpilotti: i Have that
15:11:49 <primeministerp> alexpilotti: at least some of it
15:11:50 <alexpilotti> pnavarro: I'd like to cross check as well so we can file up the relevant bugs
15:11:50 <primeministerp> one sec
15:12:00 <pnavarro> from glance, xattr and pysendfile are incompatible
15:12:40 <primeministerp> so
15:12:43 <primeministerp> this is whyat i have
15:12:54 <primeministerp> glance 2012.2
15:13:02 <primeministerp> anyjson 0.3.2+
15:13:18 <alexpilotti> pnavarro: pysendfile and xattr are dependencies of another package I guess. Glance itself?
15:13:18 <primeministerp> python-novaclient-2.6.0
15:13:54 <pnavarro> alexpilotti: yes, they are dependencies introduced in glance Folsom
15:14:05 <primeministerp> pnavarro: then we need to file an upstream bug asap
15:14:14 <primeministerp> pnavarro: do you have enough data to file a bug
15:14:31 <pnavarro> primeministerp: but, it can be considered as a bug?
15:14:45 <primeministerp> pnavarro: well we need to make someone aware
15:14:51 <primeministerp> pnavarro: or we need to fix it
15:15:45 <primeministerp> pnavarro: do we need to include new deps
15:16:14 <pnavarro> primeministerp: what it's clear, they need to be considering windows when introducing new deps
15:16:22 <primeministerp> pnavarro: correct
15:16:28 <primeministerp> jaypipes: ping
15:16:59 <jaypipes> primeministerp: pon
15:17:01 <jaypipes> g
15:17:02 <pnavarro> primeministerp: if filing a bug it's the way... why not...
15:17:16 <alexpilotti> pnavarro: I'm going to do a full folsom setup so that we can compare the results
15:17:20 <primeministerp> jaypipes: is filing a bug the best way to get windows considered when creating deps
15:17:26 <primeministerp> er introducing deps
15:17:28 <primeministerp> for example
15:17:32 <alexpilotti> pnavarro: xattr afaik is a Linux specific package
15:17:43 <primeministerp> glance now requires xattr
15:17:43 <alexpilotti> pnavarro: *nix non linux
15:17:54 <primeministerp> which is a *nix only package
15:18:05 <primeministerp> and breaks hyper-v integration
15:18:15 <jaypipes> primeministerp: glance does NOT require xattr. I specifiically removed the xattr requirement so that the image cache could work on windows.'
15:18:22 <primeministerp> o sorry
15:18:25 <alexpilotti> pnavarro: if that's the case, we have to get it out of Glance, starting w a temporary patch workaround
15:18:34 <primeministerp> jaypipes: something is requiring xattr
15:18:59 <primeministerp> jaypipes: is that in the current versions?
15:19:13 <primeministerp> jaypipes: maybe we haven't tried w/ the newer stuff since then
15:19:14 <pnavarro> jaypipes: i found it in pip-requires in F2 tar.gz
15:19:23 <jaypipes> primeministerp: well, I'm positive xattr is not needed :) perhaps this is a packaging thing? maybe xattr has been left in the packaging deps by accident.
15:19:46 <jaypipes> primeministerp: you could certainly file a bug to remove it from pip-requires (and put it in test-requires)
15:20:24 <primeministerp> pnavarro: can you add your data to a new bug re: ^^^^
15:20:32 <jaypipes> primeministerp: best to file a bug, sure, and assign it to me.
15:20:52 <primeministerp> jaypipes: thanks!!
15:20:53 <jaypipes> primeministerp: I'll move it from pip-requires to test-requires, but I'll bring mtaylor in to ensure it's not listed in any package deps.
15:20:55 <pnavarro> jaypipes: and what about pysendfile, it's used?
15:21:09 <jaypipes> pnavarro: not a requirement, same as xattr..
15:21:21 <jaypipes> pnavarro: it's used if available, ignored if not.
15:21:36 <alexpilotti> pnavarro: were you using the installer for setting up the environment?
15:22:00 <primeministerp> pnavarro: can you cc me on the bug as well
15:22:20 <pnavarro> primeministerp: ok !
15:22:33 <primeministerp> pnavarro: perfect
15:22:36 <primeministerp> pnavarro: thx
15:22:59 <primeministerp> jaypipes: ok
15:23:01 <primeministerp> er
15:23:02 <primeministerp> ok
15:23:04 <pnavarro> alexpilotti: That happened when jordan installer version was installing Folsom versions of glance
15:23:40 <alexpilotti> jaypipes: what version of Glance should we use for Folsom?
15:24:23 <jaypipes> alexpilotti: Folsom :)
15:24:29 <mtaylor> jaypipes, primeministerp: I am not the right person for removing it from pacakge deps
15:24:35 <alexpilotti> jaypipes: lol
15:24:40 <mtaylor> but I support the moving of it to test-requires
15:24:41 <jaypipes> mtaylor: zul?
15:24:46 <mtaylor> jaypipes: yah
15:24:50 <jaypipes> mtaylor: gotcha. will do.
15:24:53 <mtaylor> jaypipes: and whoever does stuff for redhat
15:25:00 <primeministerp> ayoung: ping
15:25:08 <ayoung> hey primeministerp
15:25:32 <zul> mtaylor: hmm?
15:25:34 <primeministerp> jaypipes: we need someone from rh in the discussion for removing it from package deps?
15:25:55 <ayoung> primeministerp, is this a nova thing?  markmc is your man
15:26:00 <primeministerp> ayoung: thx
15:26:27 <primeministerp> mtaylor: is zul the person to talk to re: the package deps?
15:26:44 <jaypipes> primeministerp: yeah, but I'll take care of that if you assign me the bug
15:26:49 <primeministerp> jaypipes: perfect
15:27:06 <primeministerp> ok
15:27:09 <primeministerp> we'll move on
15:27:18 <jaypipes> cheerio
15:27:23 <primeministerp> let's note bug to be filed, assigned to jaypipes
15:27:57 <primeministerp> #topic hyper-v moc interface
15:28:23 <pnavarro> jaypipes: what's your launchpad ID ?
15:28:33 <primeministerp> pnavarro: should be jaypipes
15:28:35 <primeministerp> i believe
15:29:14 <primeministerp> pnavarro: if you paste the link to the bug I can add it to the meeting notes
15:29:19 <jaypipes> yup
15:29:33 <primeministerp> jaypipes: launchpad id for pnavarro
15:29:37 <primeministerp> to add you to the bug
15:29:45 <pnavarro> https://bugs.launchpad.net/glance/+bug/1031396
15:29:47 <uvirtbot> Launchpad bug 1031396 in glance "xattr and pysendfile dependencies broke Hyper-V integration" [Undecided,New]
15:30:05 <primeministerp> #link https://bugs.launchpad.net/glance/+bug/1031396
15:30:28 <primeministerp> #topic hyper-v/chef
15:30:47 <mattray> :)
15:30:48 <primeministerp> ok let's dicuss work to automate the provisioning and prep for a hyper-v compute node
15:31:13 <primeministerp> I have implemented the chef infrastructure in my lab and began the process of gathering
15:31:30 <primeministerp> all the windows chef recipes we can use as a base for our hyper-v efforts and ci integration
15:32:18 <primeministerp> so
15:32:33 <primeministerp> mattray: is from opscode and he has offered to help us in this process
15:33:00 <mattray> ok, let me level set real quick
15:33:08 <primeministerp> mattray: yes
15:33:12 <primeministerp> mattray: please do
15:33:14 <pnavarro> jaypipes: I couldn't change the assignee of the bug to you
15:33:27 <primeministerp> mattray: i was thinking from a guidence perspective
15:33:35 <mattray> opscode.com/openstack describes the Chef for OpenStack project and links to the ML and IRC channel
15:33:54 <mattray> we have a WIP for deploying openstack with chef based on Rackspace's cookbooks
15:34:13 <primeministerp> mattray: are those specific to essex?
15:34:15 <mattray> lots of folks are working on it, the key is that it's very modular and Chef already has good support for Windows
15:34:23 <mattray> currently, Folsom work probably starts next week
15:34:32 <primeministerp> mattray: exactly i'm using the existing cookbooks as a base
15:34:59 <primeministerp> mattray: execelent, our work is specifically for configuring a hyper-v node to work w/ our openstack compute driver
15:35:04 <mattray> there is a nova/compute recipe that I plan on pulling all the kvm support out and into another cookbook
15:35:16 <primeministerp> mattray: interesting
15:35:31 <mattray> once that supports pluggable back-ends, hyper-v will be a reasonable backend
15:35:37 <mattray> (initially it will be lxc and kvm)
15:35:42 <mattray> just following the same patterns
15:35:46 <primeministerp> mattray: execelent
15:35:57 <mattray> you also need glance-client right?
15:36:05 <mattray> and keystone-client?
15:36:27 <primeministerp> mattray: no
15:36:39 <primeministerp> mattray: we only install nova
15:36:49 <alexpilotti> mattray: no, we don't
15:37:15 <pnavarro> mattray: opscode openstack cookbooks are the same that crowbar ones?
15:37:22 <mattray> pnavarro: no
15:37:29 <mattray> pnavarro: forked off awhile back by Rackspace
15:37:35 <mattray> Dell is coming back to these eventually
15:37:41 <alexpilotti> pnavarro: I just did a pip install glance on a clean VM
15:37:54 <primeministerp> mattray: then we should coordinate
15:37:55 <alexpilotti> pnavarro: I can confirm that I had the same error you had
15:38:02 <mattray> primeministerp: definitely.
15:38:07 <alexpilotti> pnavarro: there's an xattr dependency
15:38:09 <pnavarro> alexpilotti: thanks to confirm !
15:38:10 <primeministerp> mattray: so when i add our building out of nova-compute
15:38:10 <jaypipes> pnavarro: I'll assign myself...
15:38:35 <pnavarro> jaypipes: ok, thanks !
15:38:43 <alexpilotti> jaypipes: I did pip install glance, I confirm the xattr dep
15:39:18 <primeministerp> mattray: i can hopefull use some of the upstream bits as well
15:39:22 <primeministerp> mattray: specific to nova
15:39:23 <alexpilotti> jaypipes: for Folsom, should we get Glance from its github's active branch?
15:39:43 <alexpilotti> jaypipes: or can we relay on pip?
15:39:51 <jaypipes> pnavarro: done
15:40:08 <jaypipes> alexpilotti: uhm... the client or the server?
15:40:12 <mattray> primeministerp: sure. The loose plan is that sometime next week we'll start coordinating on folsom deployments and source builds. This will lend itself well to hyper-v support
15:40:27 <alexpilotti> jaypipes: client, running on hyper-v's nova-compute nodes
15:40:35 <jaypipes> alexpilotti: from PIP.
15:40:39 <primeministerp> mattray: are you expecting to have a nova chef deployment be agnostic
15:40:51 <primeministerp> i.e.
15:41:00 <alexpilotti> jaypipes: ok, we need a quick workaround to get rid of xattr
15:41:05 <primeministerp> i just have a nova recipe and it is already platform independent
15:41:49 <jaypipes> alexpilotti: yup, onit.
15:41:58 <alexpilotti> jaypipes: tx!
15:42:21 <mattray> primeministerp: I think we can work to accommodate deploying on Windows, either by attributing-out anything platform specific or with a separate recipe
15:42:32 <primeministerp> mattray: ok then
15:42:42 <primeministerp> mattray: i was going to build a seperate recipe
15:42:46 <mattray> primeministerp: there will be a lot of recipes, we'll just have to have good documentation and roles to accomodate that
15:42:47 <primeministerp> mattray: i'll continue on that path
15:43:07 <primeministerp> mattray: hopefully I can reuse a lot of what's there and transliate it to windows
15:43:13 <mattray> yeah
15:43:16 <primeministerp> mattray: when you typicall deploy
15:43:26 <primeministerp> mattray: are you doing it from package or source?
15:43:31 <primeministerp> mattray: or both
15:43:44 <mattray> primeministerp: currently packages. We're working on a strategy for source
15:44:00 <primeministerp> mattray: and for python deps, pip or also packages?
15:44:05 <mattray> all the above
15:44:13 <mattray> you show up with the code, we'll find a way to accomodate
15:44:15 <primeministerp> mattray: perfect
15:44:17 <primeministerp> hehe
15:44:19 <primeministerp> ok
15:44:29 <primeministerp> mattray: i'm going to need help running it
15:44:31 <mattray> as long as other stuff doesn't get broken, we'll kick to fit
15:44:39 <primeministerp> mattray: at least from a understanding how to push to all nodes
15:44:47 <mattray> yeah, #openstack-chef is where we gather
15:45:02 <primeministerp> mattray: ok i'll hop in there
15:45:07 <mattray> and I'm usually online, and #chef is full of 400+ helpful people
15:45:16 <primeministerp> mattray: perfect
15:45:43 <primeministerp> on that note
15:45:55 <primeministerp> let's quickly go through the bits that need to be set for windows
15:46:04 <primeministerp> 1. disable all fw's
15:46:10 <primeministerp> 2. enable iscsi target svs
15:46:16 <primeministerp> 3. set ntp
15:46:26 <primeministerp> 4. set diskpart std policy
15:46:34 <alexpilotti> jaypipes: by looking at the install log, I notice that it has also "greenlet" as a dep
15:46:42 <primeministerp> 5. live migration enablement
15:46:47 <primeministerp> 6. enable powershell
15:46:54 <alexpilotti> jaypipes: building green let fails, I'm adding a comment to the bug
15:47:04 <primeministerp> pnavarro: anything else to add to that list
15:47:18 <pnavarro> primeministerp: it's ok for me
15:47:18 <primeministerp> 7. enable hyper-v role if server is not hyper-v server
15:47:42 <primeministerp> 8. configure bridge interfaces
15:48:07 <primeministerp> those are the specific for hyper-v
15:48:08 <alexpilotti> pnavarro: I noticed that compiling green let fails not (or not yet) xattr
15:48:19 <primeministerp> for nova, we have the following
15:48:24 <alexpilotti> pnavarro: *greenlet
15:48:25 <primeministerp> that needs to be addressed by chef
15:48:42 <primeministerp> 1. installation of python
15:48:54 <primeministerp> 2. pip install of python deps
15:49:02 <primeministerp> 3. install hv code from src
15:49:11 <primeministerp> also
15:49:15 <primeministerp> 1 install python
15:49:27 <mattray> 1 & 2 sound like we need to update the python cookbook for windows support
15:49:34 <mattray> #3 will probably use the git resource
15:49:40 <primeministerp> 2 run installation from the openstack hyper-v installer
15:49:47 <primeministerp> mattray: i have one
15:49:50 <primeministerp> mattray: that i found
15:49:58 <primeministerp> mattray: i was planning on updating it today
15:50:08 <mattray> primeministerp: cool
15:50:14 <primeministerp> mattray: we use 2.7.x
15:50:22 <mattray> feel free to follow up with me directly and I'll help get your patches upstreamed quickly
15:51:02 <primeministerp> mattray: if i'm going send it upstream and I have to sign the contrib, it could take time
15:51:15 <primeministerp> i'll have to get that into lcs
15:51:16 <mattray> yeah, we need to get on that ASAP
15:51:17 <primeministerp> er lca
15:51:25 <primeministerp> mattray: do you have the link
15:51:32 <mattray> primeministerp: wiki.opscode.com/display/chef/How+to+Contribute
15:52:23 <primeministerp> mattray: how different is it than the openstack agreement
15:52:31 <mattray> it's the Apache CLA
15:52:41 <primeministerp> standard
15:52:42 <mattray> s/Apache Foundation/Opscode, Inc./
15:52:42 <primeministerp> ?
15:52:47 <primeministerp> ok
15:53:03 <mattray> it got through VMware, Dell & HP so it's not impossible ;)
15:53:26 <primeministerp> mattray: ok then, I'll start that process today
15:53:52 <primeministerp> ok
15:54:08 <primeministerp> we'll I guess we can use that as a base
15:54:19 <primeministerp> I'll send out a email to the list as soon as my starting point is online
15:54:41 <primeministerp> mattray: i'll send that to lca today
15:54:42 <mattray> ok, I'll be sure to follow up on the posts outlining how we'll be collaborating
15:54:51 <primeministerp> perfect
15:55:14 <primeministerp> ok let's see what's left to discuss
15:56:30 <primeministerp> #topic Status of Hyper-V ci Infrastructure
15:56:36 <primeministerp> so
15:56:40 <primeministerp> good news
15:57:05 <primeministerp> I have been given an asa and have configured it to support the circuit i am waiting to be pulled to my rack
15:57:45 <primeministerp> once that is done i will have control of the network infrastructure and can begin building out the pxe infrastructure to support the deployment of the hyper-v nodes
15:58:00 <primeministerp> additionally i have provisioned vpn access throught the asa
15:58:06 <primeministerp> for individuals to debug
15:58:13 <pnavarro> primeministerp: that's great !
15:58:28 <primeministerp> additional work will need to be done to get a centralized auth infrasturcutre and other long term bits
15:58:31 <primeministerp> however
15:58:32 <primeministerp> as a start
15:58:37 <primeministerp> we have some something secured
15:58:40 <primeministerp> to use
15:59:21 <primeministerp> once I have this in place I will also be able to complete tie into gerrit, as I currenlty have no outbound ssh
15:59:59 <primeministerp> anything anyone else want to add?
16:00:33 <primeministerp> ok
16:01:54 <primeministerp> ok
16:02:04 <primeministerp> if that it i'm ending the meeting
16:02:06 <primeministerp> thanks everyone
16:02:10 <primeministerp> #endmeeting