Hi Alexander, 
I traced the code and tried it out and you can override what is defined in the flavor by providing the details in the 'Extra user data' field on the launch form. For example:

filesystem_templates:
 - name: galaxy
   roles: galaxyTools,galaxyData
   archive_url: http://s3.amazonaws.com/cloudman/fs-archives/galaxyFS-dev-latest.tar.gz
   type: archive
   size: 10
 - name: galaxyIndices   
   roles: galaxyIndices
   snap_id: snap-4b20f451

If you want to use your custom bucket (i.e., if you have modified CloudMan code in some way), you can then specify it in the 'default bucket' form field and place cm.tar.gz and cm_boot.py in that bucket but no need to provide snaps.yaml. 
If you do not want to use a custom bucket, you can use the current 'Standard' flavor, which defaults to cloudman-test bucket with CloudMan from June 19th (I will not be updating anything about it until after GCC).

Hope this helps and let me know if you have more questions,
Enis

On Thu, Jul 2, 2015 at 9:29 AM, Enis Afgan <enis.afgan@irb.hr> wrote:
Hi Alexander, 
Let me investigate this a bit more and get back to you because I was under the impression I had fixed this. In the mean time, I updated your flavor to specify snap-6f0b211d.



On Wed, Jul 1, 2015 at 4:44 PM, Alexander Vowinkel <vowinkel.alexander@gmail.com> wrote:
Hi Enis,

I am trying to launch my cloudman instance with this launcher again:

You created the flavor for me because it wouldn't load my
galaxyIndices from the snaps.yaml in my default bucket.

I thought, this problem was fixed, but it still creates a galaxyIndices
from the default snapshot (600GB) when using the Standard flavor.

When I use my flavor, it loads the snapshot, but it is a little outdated ;)

So my question: Is this behaviour expected? If no, are you going to
fix it soon? If not, could you update my flavor in the way that
galaxyIndices gets created from snap-6f0b211d ?

Further:
Should I update CM? I am still on the version from Jun 22.
I am still not sure what I need to change (my snaps.yaml) in order
to have my clean cluster creation still functioning.

Thanks,
Alexander