Hi Sanjiva, Packer allows you to control most aspects of image creation, but AFAIK rely on automating the distro's installation mechanism (e.g. Debian preseed or red hat kick start). During the installation you start the guest OS so you can add and remove packages as you please. For example, on the cloudstack image, I scripted replacing the standard kernel with the xen kernel packages, and rebooted during the install to pick up the new kernel to perform other xen dependent setup. When I ran into issues early on with xen, I was considering building the kernel from source. There is a lot of flexibility.
Vagrant and packer also work with docker although I haven't tried that yet. In terms of Stratos, I have a vision forming that using vagrant and packer we could build images for different environments (aws, virtual box, docker, etc) and orchestrate the setup of stratos in those environments using vagrant - with the potential for reusing the stratos setup scripts between those environments. However, one step at a time, starting with vagrant to setup cloudstack+stratos. Cheers, Chris On 17 Mar 2014 00:47, "Sanjiva Weerawarana" <[email protected]> wrote: > > Interesting ... it would be cool to combine these with something like rPath to build a minimal Linux image with just the exact bits in it. Looks like rPath is gone - what's the way people build custom images now? Or is that gone with just more memory and more disk being normal?? > > > On Sun, Mar 16, 2014 at 10:34 PM, chris snow <[email protected]> wrote: >> >> Hi Sanjiva, >> >> Vagrant works on top of an existing image (box). There are plenty of boxes for vagrant. Ubuntu for example provides vagrant boxes [1], although the disk size is a little small to be useful. Opscode also provide some pretty good boxes [2]. If these still don't meet your needs, you can copy the packer definitions from opscode and modify them to build your own box from scratch [3]. Packer is definitely worth a look too. >> >> Many thanks, >> >> Chris >> >> --- >> [1] http://cloud-images.ubuntu.com/vagrant/ >> [2] https://github.com/opscode/bento >> [3] https://github.com/opscode/bento/tree/master/packer >> [4] http://packer.io >> >> On 16 Mar 2014 15:27, "Sanjiva Weerawarana" <[email protected]> wrote: >>> >>> Hey Chris - that's awesome .. totally +1 for having vagrant scripts as well! >>> >>> I'm not familiar with vagrant - just checking it out. Does it build a VM image or does it set up the environment to run the image? >>> >>> Sanjiva. >>> >>> >>> On Sat, Mar 15, 2014 at 1:06 PM, chris snow <[email protected]> wrote: >>>> >>>> Hi Sanjiva, >>>> >>>> For VM images, vagrant makes life very easy for users; setting up disks, setting up network cards, setting up memory, configuring guest proxy settings, running provisioning scripts, etc. >>>> >>>> I am working on a vagrant setup of cloudstack + Stratos. My project is here [1]. It isn't ready for general use yet, but I'm making good progress. Although my scripts are buggy, with a few commands I can checkout, build and provision a cloudstack developer environment. I am now working on the scripts to do the same for Stratos. >>>> >>>> Initially, the memory requirements will be high on my environment, but for me the first goal is automation, the next goal will be efficiency. >>>> >>>> Many thanks, >>>> Chris >>>> >>>> --- >>>> [1] https://github.com/snowch/devcloud-script >>>> >>>> On 15 Mar 2014 06:24, "Sanjiva Weerawarana" <[email protected]> wrote: >>>>> >>>>> I think right now we need to focus on getting a single trivial server mechanism to be able to run Stratos without too much of stuff having to be set up. I'd love to see two developer distros: >>>>> >>>>> - a VM image that has everything in it and runs in under 4GB with OpenStack + Docker. It doesn't matter whether this uses one Carbon server to run it all or whether we use RabbitMQ or other AMQP broker. (Carbon stuff HAS to run in one server - else its a bug in those products .. so the decision should not be based on ability to run in one JVM but rather just making it dirt simple to use.) This distro needs to be in 4.0.0 - I think we're nearly there for it. >>>>> >>>>> - next is a "no-IaaS-IaaS" based distro. That, we write a direct plugin to jClouds that spins up Docker images as processes and there's one JVM that works as the SM+CC+LB+AS+all. Thus the download becomes one JVM plus a URL to a Docker image registry from which the images are booted up and run (obviously a local registry will do better). We don't have this no-IaaS-IaaS yet so this can come maybe as 4.1.0 or whatever (its not that hard to make it work). >>>>> >>>>> For production deployments obviously this one server stuff is nonsense .. so we need to have full decoupled distributed execution. For that we should ship puppet scripts to get them up and running plus maybe Boto scripts for someone to get it all up on EC2 with one command. Again its totally fine to use whatever broker here and whatever other pluggable components people want to use (and we need to make sure all the parts are pluggable: load balancers, message broker, the CEP engine, etc.)). >>>>> >>>>> Makes sense? >>>>> >>>>> Sanjiva. >>>>> >>>>> >>>>> On Fri, Mar 14, 2014 at 7:59 PM, Pradeep Fernando <[email protected]> wrote: >>>>>> >>>>>> Hi Chris, >>>>>> >>>>>> Yes good point. Other day Azeez did the same suggestion. >>>>>> >>>>>> Thanks >>>>>> >>>>>> --Pradeep >>>>>> sent from my phone >>>>>> >>>>>> On Mar 14, 2014 3:47 PM, "chris snow" <[email protected]> wrote: >>>>>>> >>>>>>> Hi Pradeep - I don't know enough about how the profiles work to have a >>>>>>> view on that :( >>>>>>> >>>>>>> One thing I'm wondering though is how much memory will be saved if we >>>>>>> use RabbitMQ (or another MQ) instead of MB? >>>>>>> >>>>>>> On Fri, Mar 14, 2014 at 10:10 AM, Pradeep Fernando < [email protected]> wrote: >>>>>>> > btw, >>>>>>> > >>>>>>> > Now im working on MB and CEP bits. >>>>>>> > >>>>>>> > IMHO, we should not create MB and CEP only profiles in stratos. However, >>>>>>> > adding MB/CEP features (the ones that we use) to default profile (the >>>>>>> > profile that has all) makes sense. >>>>>>> > >>>>>>> > WDYT? >>>>>>> > >>>>>>> > Are we all on same page.. >>>>>>> > >>>>>>> > thanks >>>>>>> > >>>>>>> > >>>>>>> > On Fri, Mar 14, 2014 at 3:32 PM, chris snow <[email protected]> wrote: >>>>>>> >> >>>>>>> >> Hey Pradeep - this is exciting stuff! Looking forward to your findings! >>>>>>> >> >>>>>>> >> On Wed, Mar 12, 2014 at 5:31 PM, Pradeep Fernando < [email protected]> >>>>>>> >> wrote: >>>>>>> >> > Hi Guys, >>>>>>> >> > >>>>>>> >> > I started on the $subject. This thread is to track the progress.. >>>>>>> >> > >>>>>>> >> > thanks, >>>>>>> >> > >>>>>>> >> > >>>>>>> >> > -- >>>>>>> >> > Pradeep Fernando. >>>>>>> >> > http://pradeepfernando.blogspot.com/ >>>>>>> >> >>>>>>> >> >>>>>>> >> >>>>>>> >> -- >>>>>>> >> Check out my professional profile and connect with me on LinkedIn. >>>>>>> >> http://lnkd.in/cw5k69 >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> > -- >>>>>>> > Pradeep Fernando. >>>>>>> > http://pradeepfernando.blogspot.com/ >>>>>>> >>>>>>> >>>>>>> >>>>>>> -- >>>>>>> Check out my professional profile and connect with me on LinkedIn. >>>>>>> http://lnkd.in/cw5k69 >>>>> >>>>> >>>>> >>>>> >>>>> -- >>>>> Sanjiva Weerawarana, Ph.D. >>>>> Founder, Chairman & CEO; WSO2, Inc.; http://wso2.com/ >>>>> email: [email protected]; office: (+1 650 745 4499 | +94 11 214 5345) x5700; cell: +94 77 787 6880 | +1 408 466 5099; voip: +1 650 265 8311 >>>>> blog: http://sanjiva.weerawarana.org/; twitter: @sanjiva >>>>> Lean . Enterprise . Middleware >>> >>> >>> >>> >>> -- >>> Sanjiva Weerawarana, Ph.D. >>> Founder, Chairman & CEO; WSO2, Inc.; http://wso2.com/ >>> email: [email protected]; office: (+1 650 745 4499 | +94 11 214 5345) x5700; cell: +94 77 787 6880 | +1 408 466 5099; voip: +1 650 265 8311 >>> blog: http://sanjiva.weerawarana.org/; twitter: @sanjiva >>> Lean . Enterprise . Middleware > > > > > -- > Sanjiva Weerawarana, Ph.D. > Founder, Chairman & CEO; WSO2, Inc.; http://wso2.com/ > email: [email protected]; office: (+1 650 745 4499 | +94 11 214 5345) x5700; cell: +94 77 787 6880 | +1 408 466 5099; voip: +1 650 265 8311 > blog: http://sanjiva.weerawarana.org/; twitter: @sanjiva > Lean . Enterprise . Middleware
