[01:36] <tcliam_> hey what is the best way to track your usage on ec2?
[01:40] <flaccid> rightscale has a pretty good usage estimate system
[01:54] <tcliam_> flaccid: I am wanting to create an ebs instance for some testing but I want to stop it when I am not testing. How much will I have to pay if the ebs instance is stopped?
[01:57] <flaccid> iirc you pay the same in terms of the ebs storage allocation because it is still allocated
[01:57] <flaccid> but i could be wrong
[02:00] <Nolar> yes, you pay for ebs storage, but not ec2 runtime
[02:00] <flaccid> erichammond might know this
[02:00] <flaccid> cool
[02:00] <erichammond> know what?
 flaccid: I am wanting to create an ebs instance for some testing but I want to stop it when I am not testing. How much will I have to pay if the ebs instance is stopped?
[02:01] <flaccid> flaccid: iirc you pay the same in terms of the ebs storage allocation because it is still allocated
[02:01] <erichammond> yes
[02:02] <erichammond> $0.10-$0.11/GB/mo, depending on the region
[02:02] <erichammond> plus snapshots if you  took any
[02:04] <tcliam_> erichammond: so if I create one using an AMI from here http://uec-images.ubuntu.com/releases/10.04/release/ in terms of storage size will these AMI's have a predefined disk size or does the ebs just dynamically resize as you put data on to it?
[02:05] <Nolar> afaik the ebs amis default to 15GB volume
[02:06] <Nolar> erichammond might know better, but afaik amazon charges you for actually allocated disk on their ebs backend
[02:06] <tcliam_> Nolar: so a default install will mean paying for 15gb per month?
[02:07] <Nolar> so even though the ami would have a 15gb root, it may not actually have allocated the full 15 on the ebs backend
[02:07] <Nolar> ya, aka $1.50/mo :)
[02:07]  * erichammond nods
[02:07] <tcliam_> ah ok
[02:08] <tcliam_> and the EBS size can be extended?
[02:08] <erichammond> http://alestic.com/2009/12/ec2-ebs-boot-resize
[02:08] <erichammond> http://alestic.com/2010/02/ec2-resize-running-ebs-root
[02:09] <tcliam_> erichammond: thank you.
[02:09] <Nolar> in my case, a fresh instance only uses 923M of space out of the 15GB, so I *think* i'd only get charged 10c a month if i stopped it
[02:10] <Nolar> i've never actually tested that though :)
[02:11] <tcliam_> so if the instance is not being using i.e. not requests is there even any point in stopping it to save money?
[02:11] <tcliam_> no requests*
[02:13] <Nolar> well, stopped = no instance use hourly charges
[02:14] <Nolar> i know some folks who leave their instances stopped most of the day, but turn it back on for occasional jobs
[02:15] <Nolar> saves having to re-provision
[02:23] <tcliam_> so I just started my instance. Do I need to create an Elastic IP?
[02:23] <flaccid> no
[02:24] <tcliam_> flaccid: what is the Elastic IP for?
[02:24] <flaccid> you tell me
[02:25] <tcliam_> flaccid: assign a static IP to your instance?
[02:25] <flaccid> oh you don't know what an EIP is ?
[02:25] <tcliam_> no lol
[02:26] <flaccid> its technically not static
[02:26] <flaccid> tcliam_: http://aws.amazon.com/amis/1346
[02:28] <tcliam_> thanks
[02:28] <flaccid> np
[02:29] <Nolar> it's 'elastic' ;)
[02:29] <flaccid> thats what they call it
[02:29] <flaccid> its technically a dynamic ip
[02:29] <Nolar> sorta
[02:29] <flaccid> thats what it is in the registry
[02:29] <Nolar> it's a static address that can move dynamically between instances :)
[02:30] <flaccid> technically every address is static, i.e. the IP address doesn't change
[02:30] <Nolar> true
[02:30] <flaccid> the point is its a dynamic IP, its not marked as static in the global ip registry and they simply route it
[02:34] <Nolar> i think the point is, it's an ip which you can re-use as you change backend instances
[02:34] <Nolar> whereas the normal address instances get are fixed to only that one instance
[02:34] <Nolar> new instance, new ip
[02:34] <flaccid> thats the elastic part. i was clarifying the point that its not static which is what tcliam_  asked
[02:35] <tcliam_> ok
[02:35] <tcliam_> I'm getting this when trying to ssh into the instance " Permission denied (publickey)."
[02:36] <flaccid> tcliam_: make sure you are using an image that supports ssh and are using the correct ssh key
[02:37] <tcliam_> flaccid: is it the keypair or the private key?
[02:37] <flaccid> you use the private key or both
[02:37] <flaccid> public key needs to be in ssh on the instance, which is usually done in rc4/boot
[02:42] <tcliam_> flaccid: this is how I am trying to connect to it "ssh -i keypair.pem \ ubuntu@instancedns.com" is this correct?
[02:48] <tcliam_> aha it was the "\" that was causing the prob
[02:51] <flaccid> rightio
[02:52] <tcliam_> flaccid: so just to clarify if I want to assign a domain name to my cloud instance I must get an elastic IP point the dns record for the domain name to the elastic IP and them in my ec2 settings route the domain name to the relevant instance?
[02:52] <tcliam_> then*
[02:52] <flaccid> pretty much
[02:53] <flaccid> except the last part is called elastic ip attachment
[02:55] <tcliam_> ok
[02:55] <tcliam_> so you pay for the elastic ip by how many requests it recieves?
[02:56] <tcliam_> receives*
[03:01] <flaccid> no
[03:01] <flaccid> see the ec2 pricing
[03:02] <tcliam_> No cost for Elastic IP addresses while in use ??
[03:02] <flaccid> correct
[03:03] <tcliam_> so its free...
[03:04] <flaccid> actually i'm wrong
[03:04] <flaccid> see this http://aws.amazon.com/ec2/pricing/
[03:04] <tcliam_> yes thats what I was looking at but I dont quite understand it...
[03:04] <tcliam_> No cost for Elastic IP addresses while in use
[03:04] <tcliam_>     * $0.01 per non-attached Elastic IP address per complete hour
[03:04] <tcliam_>     * $0.00 per Elastic IP address remap – first 100 remaps / month
[03:04] <tcliam_>     * $0.10 per Elastic IP address remap – additional remap / month over 100
[03:05] <flaccid> well actually i'm right, because you pay for data transfer on any public ip, elastic or not
[09:36] <nIMBVS> hello
[09:37] <nIMBVS> anybody know where I can find some documentation regarding cloud-init's usage and configuration?
[09:38] <nIMBVS> is this it: https://help.ubuntu.com/community/CloudInit ?
[09:39] <nIMBVS> is cloud-init only configured through user-data scripts?
[09:40] <nIMBVS> or can it be configured inside the instance as well (in the case I want to create a customized public AMI)?
[09:41] <kim0> nIMBVS: hey
[09:41] <nIMBVS> hi
[09:41] <kim0> nIMBVS: no you don't need a custom ami
[09:41] <kim0> cloud-init reads the configs over the user data indeed
[09:42] <kim0> You can find extra docs at: /usr/share/doc/cloud-init/
[09:54] <nIMBVS> so all configuration for cloud-init is done through user data? what about /etc/cloud? what it's for?
[09:56] <kim0> nIMBVS: I won't claim autoritative knowledge .. but that seems to me to be the implementation of cloud-init itself
[09:56] <kim0> the configuration and templates it needs to do its work
[09:57] <kim0> nothing for you to care about ..
[09:57] <kim0> from all the examples and use cases I've seen .. you only pass info thru user data
[09:57] <kim0> which is really the use case that makes sense
[09:58] <kim0> you might want to idle around through for any of the devs to maybe correct me
[10:00] <kim0> Daviey: kirkland smoser hey folks .. The cloud forums are starting to pick up some steam
[10:00] <Daviey> kim0: awesome
[10:01] <kim0> Anyone who can answer technical questions, please follow the forums rss feed
[10:01] <kim0> like for instance http://ubuntuforums.org/showthread.php?t=1597120
[10:01] <kim0> I am trying to answer as much as I can
[10:02] <kim0> RSS Feed: http://ubuntuforums.org/external.php?type=RSS2&forumids=392
[10:02] <Daviey> kim0: "the front node seems to work fine. The nodes behind the switcher can ping between nodes but can not ping the ip outside the local network."
[10:03] <Daviey> Confuses me somewhat
[10:03] <kim0> Daviey: my understanding the cloud controller sees internet
[10:03] <kim0> the rest of the nodes .. don't
[10:03] <Daviey> kim0: Hm. that isn't what he said.
[10:03] <kim0> what's "switcher"
[10:03] <kim0> :)
[10:03] <Daviey> i assume switch
[10:04] <Daviey> but how can you be *behind* a switch
[10:04] <Daviey> surely that must mean router?
[10:04] <kim0> might just be physically
[10:04] <kim0> it seems to be two machines only
[10:05] <Daviey> Hmm.. then there should be no difference between 'front nodes' and 'nodes behind the switch'
[10:05] <Daviey> if that is what he is saying....
[10:05] <kim0> well .. if it's too unclear .. can you ask him to give more details
[10:06] <Daviey> How does this forums lark worj then :)
[10:06] <Daviey> work*
[10:06] <kim0> what lark
[10:06] <kim0> what's lark
[10:07] <kim0> Daviey: I can reply asking for more info
[10:08] <Daviey> done
[10:09] <kim0> Daviey is awesome
[10:09] <kim0> :D
[10:09] <Daviey> heh
[10:09] <Daviey> no u
[10:09] <kim0> Daviey is there a way for me to run UEC without needing multiple machines
[10:10] <kim0> I know there was this one node setup thing .. it was mostly a hack huh
[10:10] <Daviey> kim0: you really need at least 2 tbh.
[10:11] <Daviey> otherwise you are using an unsupported topology, making it less than useful
[10:11] <kim0> Daviey: can it run over ec2
[10:11] <kim0> would that be supported or close
[10:11] <Daviey> close
[10:11] <Daviey> you'd be using qemu rather than kvm
[10:12] <Daviey> so, sloooooooooooooooooooooow
[10:12] <kim0> so running vms slow as hell
[10:12] <kim0> yeah :)
[10:12] <kim0> Amazon should buy nested virt chips :)
[10:19] <nIMBVS> kim0: you're the same kim0 that made this presentation: http://irclogs.ubuntu.com/2010/10/13/%23ubuntu-classroom.html#t17:00 ?
[10:19] <kim0> nIMBVS: yeah
[10:20] <nIMBVS> cool
[10:20] <kim0> was I that bad :)
[10:20] <nIMBVS> no. it was ok. very helpful ot clarify some things
[10:20] <kim0> cool
[16:06] <kirkland> kim0: sorry, i can't keep up with the forums
[16:06] <kim0> kirkland: it's only an RSS
[16:06] <kirkland> kim0: irc, mailing lists, LP bugs, and LP answers
[16:06] <kim0> no need to poll for change :)
[16:06] <kirkland> that's all i can do
[16:06] <kirkland> kim0: sorry, someone else is going to have to cover that
[16:06] <kirkland> kim0: i'm out of bandwidth
[16:07] <kim0> ok no worries
[16:28] <nIMBVS> does anybody know what gzip-rsyncable program is?
[16:29] <nIMBVS> I didn't find it anywhere in the repository
[16:38] <nIMBVS> nevermind. I found gzip-rsyncable
[16:40] <RoAkSoAx> kirkland: in the NC do I also haveb to specify "VNET_CLOUDIP="<CLC_ipaddress>"" ?
[16:40] <RoAkSoAx> kirkland: is it relevant if I do?
[17:16] <kirkland> RoAkSoAx: yes, i think that's important
[17:36] <RoAkSoAx> kirkland: another question. When you run an instances, the 2GB image store file is stored at the Walrus right? So, what is stored in the NC?
[17:37] <kirkland> RoAkSoAx: a local cached copy of that
[17:39] <RoAkSoAx> kirkland: so what happens if the Walrus dies? will the instance die too?
[18:25] <RoAkSoAx> kirkland: since the NC is a VM, is it possible to run instances inside that VM? what will it be necessary to do that because I get this error: [Fri Oct 15 13:23:55 2010][001562][EUCAERROR ] libvirt: internal error no supported architecture for os type 'hvm' (code=1)
[18:25] <RoAkSoAx> [Fri Oct 15 13:23:55 2010][001562][EUCAFATAL ] hypervisor failed to start domain
[18:26] <kirkland> RoAkSoAx: you need to make a minor change to the libvirt xml
[18:26] <kirkland> RoAkSoAx: replace kvm with qemu for the hypervisor
[18:27] <RoAkSoAx> kirkland: thanks ;)
[18:27] <kirkland> RoAkSoAx: np;  it's going to be *really* slow
[18:28] <RoAkSoAx> kirkland: yeah I just want it to test what would happen if the CLC dies while having running instances
[18:37] <RoAkSoAx> kirkland: i cant seem to find the xml file, where is it located exactly?
[18:48] <RoAkSoAx> found it
[19:01] <RoAkSoAx> kirkland: Ok, Successfully started an image, it running pretty fast. Then I failed CLC1, CLC2 took control, and I runned a euca-describe-instances, and the instace I runned was there...
[19:02] <kirkland> RoAkSoAx: sweet ;-)
[19:03] <RoAkSoAx> kirkland: though, the instances that were previously terminated are no longer there
[19:03] <RoAkSoAx> kirkland: is that *normal* ?
[19:03] <RoAkSoAx> or even expected behaviour?
[19:15] <timwood> Afternoon.
[19:16] <timwood> Does anyone know how to increase the size of the root file system in the default Ubuntu EMI from the image store?
[19:24] <RoAkSoAx> hggdh: i think i just found a bug. I turned everything off but the CLC, and it still shows that there's a running instance
[19:25] <hggdh> RoAkSoAx: no CCs running at all?
[19:26] <RoAkSoAx> hggdh: nope not at all
[19:27] <RoAkSoAx> hggdh: here's how I discovered. I had terminated instances and 1 running instances. Then I did a failover (failed CLC1, CLC2 took control) and it showed the running instance but not the terminated ones
[19:27] <RoAkSoAx> then powered off CC
[19:27] <RoAkSoAx> still showed running instances
[19:27] <RoAkSoAx> then powered off walrus
[19:28] <RoAkSoAx> still showing running instance
[19:28] <RoAkSoAx> then powered NC off
[19:28] <RoAkSoAx> still kept on showing running instance
[19:29] <RoAkSoAx> hggdh: so this is either because of the failover or because there is a real bug there...
[19:36] <hggdh> RoAkSoAx: I am not sure either
[19:36] <hggdh> but, anyway, interesting results :-)
[19:37] <RoAkSoAx> hggdh: indeed
[19:37] <hggdh> RoAkSoAx: try bring up CC Walrus, and NC -- then see if this "running" instance is actually restarted on teh NC
[19:39] <RoAkSoAx> hggdh: will do in a bit.. i really need to get off bed xD
[21:58] <jeremydei> soren, I'm messing with building a maverick image with vm-builder .. it seems rather simple to extend to Maverick from lucid .. was there any reason this change wasn't included in the Maverick release?
[21:59] <soren> jeremydei: Not a good one, no.
[22:11] <jeremydei> soren, cool, i'll let u know how it goes :)  then maybe i'll look and see if i can do a pull-req for ya
[22:11] <jeremydei> right now just editing the files direct, and it seems to be working
[23:45] <RoAkSoAx> kirkland: quick question. the br1 in the CC is used to be able to communicate with the instances only when using --addressing private?