advice on a mixed cloud/SGE architecture
Hi, I was contemplating setting up a galaxy master node on a aws instance but executing the workflows on an internal SGE cluster. I am not yet very familiar with galaxy so wanted to ask whether this is feasible (i.e. the API would support it). Any advice/recommendation is most welcome. many thanks, Nick
Hi Nick, Not sure I completely understand what you're trying to do here. So, you want to have a galaxy head node on AWS that dispatches jobs to a non-AWS private SGE cluster? -Dannon On Thu, May 16, 2013 at 12:18 PM, Nikolay N. <nikolay12@gmail.com> wrote:
Hi,
I was contemplating setting up a galaxy master node on a aws instance but executing the workflows on an internal SGE cluster. I am not yet very familiar with galaxy so wanted to ask whether this is feasible (i.e. the API would support it). Any advice/recommendation is most welcome.
many thanks, Nick
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
Hi Dannon, Yes, this is what I was thinking about. The reason is that at the moment the SGE cluster is managed separately. Also we are likely to migrate to a different cluster infrastructure (probably LSF) and I thought that if I set up a master node on AWS I could than "only" have to change the cluster part. The idea is that if the user metadata is stored on the AWS master node and his/her files on S3 than it would be relatively easy to just "swap" the cluster that is providing the "raw muscle". But I don't have a good understanding of the galaxy architecture and the documentation seems a bit sparse. So any advice on this is most welcome. -Nick On Tue, May 28, 2013 at 3:58 PM, Dannon Baker <dannon.baker@gmail.com>wrote:
Hi Nick,
Not sure I completely understand what you're trying to do here. So, you want to have a galaxy head node on AWS that dispatches jobs to a non-AWS private SGE cluster?
-Dannon
On Thu, May 16, 2013 at 12:18 PM, Nikolay N. <nikolay12@gmail.com> wrote:
Hi,
I was contemplating setting up a galaxy master node on a aws instance but executing the workflows on an internal SGE cluster. I am not yet very familiar with galaxy so wanted to ask whether this is feasible (i.e. the API would support it). Any advice/recommendation is most welcome.
many thanks, Nick
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
Ok, got it. Unfortunately this isn't really possible at this time, at least not without a lot of hacking and dealing with data staging issues and delays (in downloading data from S3 to local cluster to work, uploading results back to galaxy server, etc.). Galaxy, with the exception of the LWR - http://wiki.galaxyproject.org/Admin/Config/LWR, relies on a shared filesystem between the galaxy instance and workers. I'd really recommend colocating Galaxy with the cluster, at least for now. We do want to build out the ability to use EC2 as an expansion to local clusters, but I imagine it working the other way around -- that is, you have a local instance that uses S3 as permanent storage and spin up extra nodes in EC2 to handle excess load as necessary. -Dannon On Tue, May 28, 2013 at 1:14 PM, Nikolay N. <nikolay12@gmail.com> wrote:
Hi Dannon,
Yes, this is what I was thinking about. The reason is that at the moment the SGE cluster is managed separately. Also we are likely to migrate to a different cluster infrastructure (probably LSF) and I thought that if I set up a master node on AWS I could than "only" have to change the cluster part.
The idea is that if the user metadata is stored on the AWS master node and his/her files on S3 than it would be relatively easy to just "swap" the cluster that is providing the "raw muscle". But I don't have a good understanding of the galaxy architecture and the documentation seems a bit sparse. So any advice on this is most welcome.
-Nick
On Tue, May 28, 2013 at 3:58 PM, Dannon Baker <dannon.baker@gmail.com>wrote:
Hi Nick,
Not sure I completely understand what you're trying to do here. So, you want to have a galaxy head node on AWS that dispatches jobs to a non-AWS private SGE cluster?
-Dannon
On Thu, May 16, 2013 at 12:18 PM, Nikolay N. <nikolay12@gmail.com> wrote:
Hi,
I was contemplating setting up a galaxy master node on a aws instance but executing the workflows on an internal SGE cluster. I am not yet very familiar with galaxy so wanted to ask whether this is feasible (i.e. the API would support it). Any advice/recommendation is most welcome.
many thanks, Nick
___________________________________________________________ Please keep all replies on the list by using "reply all" in your mail client. To manage your subscriptions to this and other Galaxy lists, please use the interface at: http://lists.bx.psu.edu/
To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/mailinglists/
participants (2)
-
Dannon Baker -
Nikolay N.