Galaxy has been highly customizable in the past. What is the proper path for deploying a customized version of Galaxy in AWS (or any cloud). I’ve struggled to find the correct balance between a bare bones installation of galaxy and something completely turn-key like GVL’s cloud launcher. I like what cloudman seems to offer but I can’t seem to find a good resource for how one might leverage cloudman while using a highly customized version of galaxy. I also get the sense that many of the guides for cloudman are now deprecated as the project has been moved into GVL.
While CloudMan is still very much a part of the Galaxy project, and is also used as the infrastructure manager in the GVL project, it has been going through a lot of change lately because it was rewritten from the ground-up to support Kubernetes. Therefore, we are no longer developing the old version of CloudMan, and the focus is on v2. You can launch alpha versions of this CloudMan version by selecting CloudMan 2 or GVL 5 alpha, which are more or less the same thing at present.
CloudMan will provision VMs on the target cloud, setup a Kubernetes cluster on it, and deploy a Kubernetes Helm Chart for Galaxy (https://github.com/galaxyproject/galaxy-helm), and therefore, customising Galaxy is a matter of customising the Helm chart.
Alternatively, you can also deploy your Galaxy Helm chart directly if you already have an existing k8s cluster and do not need CloudMan to manage it. The Galaxy Helm chart has already gone through a few iterations and we consider the interface to be fairly stable.
Thanks that helps. One other follow up question. I was able to get the cloud launcher django app running but it didn’t show any appliances. Is there a resource on how to add those. (say if I wanted to add a customized version of galaxy and cloudman)
Thanks, I followed up on your comment about using cloudman2. Does that spin up a cluster? It shows zero nodes and when I try to add a node it errors out.
By default, it just starts up a master node where all the services are running. Adding worker nodes is for expanding the capacity for user jobs. What was the error when trying to adda node? Does it happen on AWS?
When I select add node, it spins on retrieving instance types and gives the following JS error:
vendor.js:77185 ERROR TypeError: Cannot read property ‘target’ of undefined
at MapSubscriber.project (main.js:723)
at MapSubscriber.push…/node_modules/rxjs/_esm5/internal/operators/map.js.MapSubscriber._next (vendor.js:170048)
at MapSubscriber.push…/node_modules/rxjs/_esm5/internal/Subscriber.js.Subscriber.next (vendor.js:164766)
at TapSubscriber.push…/node_modules/rxjs/_esm5/internal/operators/tap.js.TapSubscriber._next (vendor.js:172781)
at TapSubscriber.push…/node_modules/rxjs/_esm5/internal/Subscriber.js.Subscriber.next (vendor.js:164766)
at TapSubscriber.push…/node_modules/rxjs/_esm5/internal/operators/tap.js.TapSubscriber._next (vendor.js:172781)
at TapSubscriber.push…/node_modules/rxjs/_esm5/internal/Subscriber.js.Subscriber.next (vendor.js:164766)
at ForkJoinSubscriber.push…/node_modules/rxjs/_esm5/internal/observable/forkJoin.js.ForkJoinSubscriber.notifyComplete (vendor.js:165935)
at InnerSubscriber.push…/node_modules/rxjs/_esm5/internal/InnerSubscriber.js.InnerSubscriber._complete (vendor.js:163986)
at InnerSubscriber.push…/node_modules/rxjs/_esm5/internal/Subscriber.js.Subscriber.complete (vendor.js:164778)
defaultErrorLogger @ vendor.js:77185
DashboardGrid.tsx:174 size is zero!