Description of problem:

After an upgrade from 5.5 to 5.6, while moving from rubyrep to pglogical the following error was encountered (in the postgresql.log):
root@vmdb_production:[14630]:ERROR:  duplicate key value violates unique constraint "index_storages_vms_on_vm_id_and_storage_id"
root@vmdb_production:[14630]:DETAIL:  Key (vm_or_template_id, storage_id)=(10000000003768, 10000000000394) already exists.
root@vmdb_production:[14630]:CONTEXT:  COPY storages_vms_and_templates, line 2604
root@vmdb_production:[14630]:STATEMENT:  COPY "public"."storages_vms_and_templates" FROM stdin
root@vmdb_production:[14630]:ERROR:  current transaction is aborted, commands ignored until end of transaction block
root@vmdb_production:[14630]:STATEMENT:  COPY "public"."container_groups_container_services" FROM stdin
@:[14271]:ERROR:  table copy failed

This indicates that a row was present in the global database from the region that should not have been there.

Upon closer investigation, the following was found in the storages_vms_and_templates table:

vmdb_production=# select * from storages_vms_and_templates where vm_or_template_id = 10000000003768 and storage_id = 10000000000394;
   storage_id   | vm_or_template_id |       id
----------------+-------------------+----------------
 10000000000394 |    10000000003768 | 99000000000015
(1 row)

This shows that this join table was assigned an id out of its proper region.
Because of this, this row was not seen by `bin/rake evm:dbsync:destroy_local_region 10` and disrupted the sync.