09:02:00 <viglesia1> #startmeeting
09:02:00 <eucabot> Meeting started Thu Mar 15 16:02:22 2012 UTC.  The chair is viglesia1. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:02:00 <eucabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
09:02:00 <viglesia1> #topic Eutester
09:02:00 <viglesia1> okay then
09:02:00 * mull lurks
09:03:00 <viglesia1> Today i wanted to ping an idea i had about restructuring some classes to make it easier to work with eucalyptus bare metal machines and the cloud together
09:03:00 <viglesia1> #topic Interacting with bare metal machines
09:03:00 <viglesia1> so as it stands today in eutester we have a bare metal machine class
09:04:00 <viglesia1> this represents the distro, distro version, components installed, and ip of the bare metal machine
09:05:00 <viglesia1> what i would like to do is create a child class of these for each euca component
09:05:00 <viglesia1> then you could have something like tester.clc.stop()
09:05:00 <viglesia1> or tester.cc.cleanrestart
09:05:00 <jeevan_ullas> nice this would help test HA and stuff.
09:05:00 <viglesia1> hahah exactly the goal
09:06:00 <viglesia1> The other thing that we have been working on with mclark is extending the boto instance class
09:06:00 <mull> yeah, you definitely have to dela with master/slave there
09:06:00 <mull> s/dela/deal/
09:07:00 <viglesia1> yeah that will all be part of it, we will have a mechanism that populates the current masters
09:07:00 <viglesia1> then from there the calls to a particular machine (ie tester.clc) will always be to the master
09:07:00 <viglesia1> any other gotchas you can think of mull?
09:08:00 <viglesia1> i know youve had some fun with this in the past
09:08:00 <mull> viglesia1, well, sometimes people will really _want_ to call the slave (to test redirects)
09:08:00 <mull> but maybe those aren't the sort of functions you're thinking of for this
09:08:00 <viglesia1> well i can make sure to keep that in mind
09:09:00 <viglesia1> hopefully i can find a clean way to access the slave when you need
09:09:00 <viglesia1> it
09:09:00 <mull> anyway, this is a good place to use a python @property
09:09:00 <viglesia1> that is a function that populates a variable right?
09:09:00 <viglesia1> if it doesnt exist?
09:09:00 <mull> right, so you can treat "clc" like an object, but you actually call a function to figure out which is master
09:10:00 <mull> and it returns the correct object each time
09:10:00 <viglesia1> i like
09:10:00 <viglesia1> great idea
09:10:00 <viglesia1> is it in 2.6?
09:10:00 <mull> yeah, been around forever
09:10:00 <viglesia1> sweet, sorry im a newb
09:10:00 <mull> :)
09:11:00 <viglesia1> k
09:12:00 <viglesia1> so another thing we are working on is a way to use instance DNS and  DNS delegation without having to setup an actual server
09:12:00 <viglesia1> anybody have ideas on that within python?
09:12:00 <viglesia1> ie setting up a dummy server on the fly or something of the sort?
09:12:00 <obino> viglesia1: do you mean for the instances IP?
09:12:00 <mull> viglesia1, dnsmasq would be the standard way, I think
09:12:00 <viglesia1> but nothing that i can use programatic in python?
09:13:00 <mull> you're talking about being able to point to sort of a "fake" DNS server, right?
09:13:00 <viglesia1> i was more wondering about the dns delegation bit
09:14:00 <viglesia1> yeah i dont know if anybody has used a particular lib for DNS in python before
09:15:00 <obino> viglesia1: why do you need the delegation part? You can test eucalyptus DNS directly right?
09:15:00 <mull> I haven't tried that.  It probably exists.  the goal is to allow your boto connections to resolve host names for services, right?
09:15:00 <viglesia1> yeah
09:15:00 <dfederlein> question: is this for f5 style load balancing or simply a masking of a single ip?
09:15:00 <viglesia1> obino: dfederlein its for HA
09:15:00 <dfederlein> right ok
09:16:00 <viglesia1> the MAster and slave should both report the master IP
09:16:00 <dfederlein> so you're looking to figure out how to basically load balance and return the active only until failover.
09:16:00 <viglesia1> right
09:16:00 <dfederlein> sorry to derail, just clarifying for my own
09:16:00 <obino> viglesia1: any linux machine should use /etc/hosts
09:16:00 <obino> you could change it manually at failover time
09:17:00 <viglesia1> yeah id like to stay away from editing files as much as possible though
09:17:00 <mull> viglesia1, perhaps dnspython is the answer here
09:17:00 <viglesia1> *looks*
09:18:00 <mull> I'm still investigating
09:18:00 <mull> not sure if it actually does what I thought it might
09:19:00 <viglesia1> guess we still need to poke around, might be able to build something particulart to this purpose
09:19:00 <viglesia1> should be pretty simple message handling
09:20:00 <viglesia1> mull thanks for the pointer
09:21:00 <viglesia1> k then those were the two things i wanted to talk about today
09:21:00 <jeevan_ullas> is there any plan to package eutester for linux distros
09:21:00 <viglesia1> i dont believe we are stable enough for that yet
09:21:00 <jeevan_ullas> ?
09:21:00 <viglesia1> i would consider it still in development
09:22:00 <jeevan_ullas> ok
09:22:00 <viglesia1> once we figure out the HA management, we'll be closer to a "release"
09:22:00 <viglesia1> for now its a huge chunk missing
09:22:00 <jeevan_ullas> i see IAM stuff in there
09:22:00 <viglesia1> yeah that is another area that needs to be bolstered
09:22:00 <viglesia1> having IAM convinience functions will be huge
09:23:00 <viglesia1> #topic open floor
09:23:00 <viglesia1> anybody else have any questions comments, answers?
09:23:00 <dfederlein> thoughts on Eucaschool for eutster?
09:24:00 <jeevan_ullas> can we use this against aws too? to verify similarites between aws and euca
09:24:00 <jeevan_ullas> like compatibility issues
09:24:00 <mull> jeevan_ullas, I think that's one of its goals
09:24:00 <viglesia1> dfederlein: yes i am going to work on it this weekend, hopefully have something solid and let you know if i am ready monday
09:24:00 <viglesia1> jeevan_ullas: yep
09:24:00 <dfederlein> ok cool.  Are you in need of mikogo or gtm?
09:25:00 <jeevan_ullas> cool. regarding that eucaschool are you guys only planning irc
09:25:00 <dfederlein> for screen examples?
09:25:00 <viglesia1> im pretty sure it would help
09:25:00 <dfederlein> or, as an alternative, a screen-shared google hangout?
09:25:00 <jeevan_ullas> ya was coming to that dfederlein
09:25:00 <viglesia1> what have you guys seen work in the past
09:25:00 <dfederlein> all of those.
09:25:00 <dfederlein> but mikogo and google are truly OS independent
09:26:00 <dfederlein> ie: anyone can join those on any os.
09:26:00 <viglesia1> i mean i feel liek its super boring to watch someone code, but watching someone type about code is worse
09:26:00 <jeevan_ullas> i have been to fedora-classroom previously and they were all irc
09:26:00 <dfederlein> lol if you think just the irc sessions would be ok, then we're good.  the only thing I was thinking was if you wanted to demonstrate something
09:26:00 <viglesia1> so would the main draw be the mikogo or the irc
09:26:00 <dfederlein> it would be irc.
09:26:00 <obino> dfederlein: viglesia1: I have seen IRC school using shared 'screen' session if it's only CLI
09:26:00 <dfederlein> you would only use the desktop sharing if you needed to show
09:27:00 <dfederlein> obino: that works too
09:27:00 <dfederlein> I'm totally open tow hatever is most effective with IRC
09:27:00 <dfederlein> *to whatever
09:27:00 <viglesia1> yeah all ill really need is a terminal i think
09:27:00 <viglesia1> will know more when i have the tutorial setup
09:27:00 <jeevan_ullas> considering the nature of complexity i dont think people can try out anything while you are giving the classroom session on irc
09:27:00 <dfederlein> ok we can have an instance on ECC or something with a screen session I guess
09:28:00 <dfederlein> ok you let me know what you need, I'll get it ready
09:28:00 <viglesia1> k
09:28:00 <viglesia1> okay guys any more q's or comments?
09:28:00 <jeevan_ullas> nope. i am good :)
09:29:00 <viglesia1> alright then
09:29:00 <dfederlein> thanks for humoring me
09:29:00 <viglesia1> thanks all for coming by, i really appreciate the input
09:29:00 <viglesia1> dfederlein: im funny like a clown funny?
09:29:00 <viglesia1> #endmeeting