nodewatcher: fetch instance type only at start-up

Since the compute node cannot change its instance type it makes no sense
to dynamically fetch this value in the nodewatcher loop.
This change reduces the calls to DescribeStack API which allows a relatively
low TPS

Signed-off-by: Francesco De Martino <fdm@amazon.com>

Author: demartinofra

nodewatcher/nodewatcher.py

@@ -296,20 +296,13 @@ def _poll_instance_status(config, scheduler_module, asg_name, hostname, instance
     _terminate_if_down(scheduler_module, config, asg_name, instance_id, INITIAL_TERMINATE_TIMEOUT)
 
     idletime = _init_idletime()
-    instance_type = None
+    instance_type = get_compute_instance_type(config.region, config.proxy_config, config.stack_name)
+    instance_properties = get_instance_properties(config.region, config.proxy_config, instance_type)
     while True:
         time.sleep(60)
         _store_idletime(idletime)
         _terminate_if_down(scheduler_module, config, asg_name, instance_id, TERMINATE_TIMEOUT)
 
-        # Get instance properties
-        new_instance_type = get_compute_instance_type(
-            config.region, config.proxy_config, config.stack_name, fallback=instance_type
-        )
-        if new_instance_type != instance_type:
-            instance_type = new_instance_type
-            instance_properties = get_instance_properties(config.region, config.proxy_config, instance_type)
-
         has_jobs = _has_jobs(scheduler_module, hostname)
         if has_jobs:
             log.info("Instance has active jobs.")