fix terminate-instances

[GoogleCodeExporter]

Previously, the user could do run-instances to run appscale a single time
and if they wanted to run it again, they could just do run-instances again.

We can dramatically improve startup time for hbase and hypertable if we use
the standard model of forcing the user to terminate then run if they want
to run a second time.

Need to move all terminate-related stuff into its own script to correctly
teardown an appscale deployment.

Original issue reported on code.google.com by shattere...@gmail.com on 2 Nov 2009 at 9:24

[GoogleCodeExporter]

In progress. terminate-instances now runs shutdown-datastore to cleanly 
shutdown the
database, and pkills if it cannot do so. tested against all databases, need to 
clean
up a little bit more for hypertable still.

run-instances now also guards against running twice in a row with the same 
keyname,
and will abort if the keyname is already in use. This behavior can be 
overridden with
the --force flag (although it is only recommended to do so for ec2/euca and not 
for
xen/kvm).

Original comment by shattere...@gmail.com on 4 Nov 2009 at 12:55

• Changed state: Started

[GoogleCodeExporter]

Done. Fixed for hypertable, so all databases are cleaned up properly now.

Original comment by shattere...@gmail.com on 4 Nov 2009 at 1:36

• Changed state: Fixed