[Arvados] Deploy Arvados

Tom Clegg tom at clinicalfuture.com
Thu Apr 25 13:07:09 EDT 2013

Hi Cornelius,

I've posted some more release status info here. Short version: Arvados
is not yet deploy-ready.


A Keep volume is just a regular filesystem (typically ext4) which Keep
(our distributed CAS) uses as its back-end store. Documentation for
setting them up is in the works, along with a rewrite of Keep itself.

Technically there is no hard requirement to run the API service (or
anything else) in a VM. However, in a production environment, the
virtualization layer is necessary to achieve the performance,
efficiency, and scalability that we are aiming for. It also tends to
make administration less painful.

The "install" page on the doc site is really just the beginning of an
outline. We will be adding much more detail in the next few months.


On Mon, Apr 22, 2013 at 12:35 AM, Bock, Cornelius
<Cornelius.Bock at student.hpi.uni-potsdam.de> wrote:
> Hello everybody,
> I'm interested in the Arvados-Project and want to deploy it. Are there any more detailed instructions than in the
> install guide<http://doc.arvados.org/install/>?
> I have some Linux boxes here which should work as a cluster. They do also have access to an in-memory database.
> I could create an nfs-share, but how do i create a "Keep volume"?
> Do i need to use a VM for the Arvados REST API server? Why can't i just run it directly? Are there more detailed
> instructions about installing this, especially about configurations i have to choose?
> How do i create an administrator account?
> What is the job manager, where should i deploy it and how?
> What else do i need?
> Thank you for any information!
> Best regards,
> Cornelius Bock
> _______________________________________________
> Arvados mailing list
> Arvados at arvados.org
> http://lists.arvados.org/mailman/listinfo/arvados

More information about the Arvados mailing list