I have set up a couple of Galaxy instances using CloudMan (c4.8xlarge, 250-1000Gb EBS volumes), and return to them periodically for more analysis. Although the instance shuts down, the EBS volumes persist, and they can easily be restarted. However, given the expense of maintaining EBS volumes, I would like to take a snapshot (S3 storage rates), and simply re-volume from that when I need it. However, this doesn't seem to work. Is there a way to do something like this so that my data can sit on S3 or Glacier until I need it?
There is no automated/supported method to do this (although work is underway within Galaxy to add this feature but that's still some time out). What you could do is to stop all services via CloudMan, manually create a tarball of /mnt/galaxy
directory and upload that to S3. You could then delete the EBS volume via the AWS dashboard. At a later point then, create a new Galaxy CloudMan instance with a new volume of at least the size the stored, uncompressed data is. After the system starts up, stop all services via CloudMan again, replace the contents of /mnt/galaxy
with the contents of the earlier created tar file from S3 and reboot the cluster for things to get reloaded.
I've never tried this but it should work pretty easily. I'd certainly try it out once or twice without deleting the EBS volume and make sure all the nuances are accounted for. Feel free to ping with more specific questions if you embark on that route and we'll figure it out.