Chad A. Musson's picture

I created a "micro" instance providing about 200gb of data for remote users.   The data is in an EBS volume attached to the instance (data only, no apps in the volume).

I'd like to detach the volume, and attach it to a different instance temporarily, then switch it back to the original instance.   I just need to increase the CPU/RAM for a one-time specific task.

I can't find a way to do this via the Hub.  

Is there an easy way to do this?

Forum: 
Jeremy Davis's picture

Its not something I've ever tried so you'll have to have a play I guess.

But with 200GB of data to lose or get corrupted the first thing I'd be doing is make sure you have a backup!

I'm just not sure how you detach the volume from your micro instance - or even if you need to? Perhaps just shutting down the micro instance will be enough? Then if you shutdown the machine, and start a new small instance and try to attach your EBS volume (the Hub allows attaching of EBS volumes).

If that doesn't work you'll need some advice from someone more experienced with it than me. :)

Alon Swartz's picture

Firstly you need to know that EBS volumes are region specific (actually zone specific but the Hub handles those specifics).

Anyway, what you need to do is:

  • Unmount the volume from with in the micro instance
  • Detach the volume from with in the Hub (there is a "detach" option in the server record next to the volume description)
  • Launch a new instance, small/medium (make sure its in the same region, e.g. Virginia (East USA)
  • Attach the EBS volume to the new instance. It should be automatically mounted to the same mountpoint as it was in the micro instance - that is if you let ebsmount do its stuff).

I hope the above helps.

Chad A. Musson's picture

Okay, the "attach existing" only shows when there are unattached EBS volumes, rather than showing all the time, with "none available".

The 200gb of data isn't the master copy, that's currently in-house.  After transferring the data, we discovered that an additonal conversion step is needed.   It's easier to run the conversion on the data already transferred, than to transfer it a second time (assuming enough CPU/RAM).

Thanks!


Add new comment