Hi, I have been reading some docs about cloudman and I am thinking to give it a try, but I have a few doubts about it. 1) I have read in the wiki that it's only available in region US East (Virginia). Will it be available in Ireland region? 2) I see I need to attach EBS volumes to my head node to host the software and data. I think EBS volumes have a max size of 1TB. Which would be the approach if I need a filesystem bigger than 1TB? 3) Is possible to customize compute nodes with extra packages or config files? I suppose once the compute nodes are up and running I could ssh to them and do any modification but if the auto-scaling feature launch new compute nodes those new lauched nodes would not have my custom modifications. Is possible to do modifications that keep persistent even for new launched compute nodes? thank you very much for your help. Pablo.
Hi Pablo, 1) I have read in the wiki that it's only available in region US East
(Virginia). Will it be available in Ireland region?
This is currently true, though we're hoping to mirror to other regions soon.
2) I see I need to attach EBS volumes to my head node to host the software and data. I think EBS volumes have a max size of 1TB. Which would be the approach if I need a filesystem bigger than 1TB?
You are correct that the maximum EBS volume size is 1TB. We have an S3 object store in development (beta-ish, though some are already using it) that will replace the primary EBS-based storage as the default cloud galaxy object store using (free) instance ephemeral storage for the lifespan of a cluster, having all datasets persisted to S3 for the long term. This is on my roadmap to do this year, if the time-frame is useful. 3) Is possible to customize compute nodes with extra packages or
config files? I suppose once the compute nodes are up and running I could ssh to them and do any modification but if the auto-scaling feature launch new compute nodes those new lauched nodes would not have my custom modifications. Is possible to do modifications that keep persistent even for new launched compute nodes?
Sure, you can do this using a worker post start script configured for your cluster. See 'worker_post_start_script_url' at https://wiki.galaxyproject.org/CloudMan/UserData.
participants (2)
-
Dannon Baker
-
Pablo Escobar