Commit 009c7887 authored by Paul Rich's avatar Paul Rich
Browse files

Fixing more issues from shakeout testing

Non-idle nodes are now fully respected.  Consistiently get string nid
lists out of this.  ValueError doesn't get raised if the attrs location
exists stradling a reservation (still in the queue, but not available
due to the reservation).
parent 3a2317cc
......@@ -685,27 +685,28 @@ class CraySystem(BaseSystem):
# we also have to forbid a bunch of locations, in this case.
idle_nodecount = 0
unavailable_nodes = []
forbidden = set(self.chain_loc_list(job.get('forbidden', [])))
required = set(self.chain_loc_list(job.get('required', [])))
requested_locations = expand_num_list(job['attrs'].get('location', ''))
forbidden = set([str(loc) for loc in self.chain_loc_list(job.get('forbidden', []))])
required = set([str(loc) for loc in self.chain_loc_list(job.get('required', []))])
requested_locations = [str(n) for n in expand_num_list(job['attrs'].get('location', ''))]
requested_loc_in_forbidden = False
for loc in requested_locations:
if loc in forbidden:
#don't spam the logs.
requested_loc_in_forbidden = True
break
if not job['queue'] in self.nodes_by_queue.keys():
# Either a new queue with no resources, or a possible
# reservation need to do extra work for a reservation
node_id_list = list(required - forbidden)
for node_id in node_id_list:
if self.nodes[str(node_id)].status in ['idle']:
idle_nodecount += 1
else:
unavailable_nodes.append(node_id)
for node_id in unavailable_nodes:
node_id_list.remove(node_id)
else:
idle_forbidden_count = len([nid for nid in forbidden
if self.nodes[str(nid)].status =='idle'])
idle_nodecount = idle_nodes_by_queue[job['queue']] - idle_forbidden_count
node_id_list = list(set(self.nodes_by_queue[job['queue']]) - forbidden)
for node_id in node_id_list: #strip non-idle nodes.
if not self.nodes[str(node_id)].status in ['idle']:
unavailable_nodes.append(node_id)
for node_id in unavailable_nodes:
node_id_list.remove(node_id)
if requested_locations != []: # handle attrs location= requests
job_set = set([int(nid) for nid in requested_locations])
job_set = set([str(nid) for nid in requested_locations])
if not job['queue'] in self.nodes_by_queue.keys():
#we're in a reservation and need to further restrict nodes.
if job_set <= set(node_id_list):
......@@ -714,10 +715,11 @@ class CraySystem(BaseSystem):
else:
# We can't run this job. Insufficent resources in this
# reservation to do so. Don't risk blocking anything.
idle_nodecount = 0
#idle_nodecount = 0
node_id_list = []
else:
if job_set <= set([int(node_id) for node_id in
#normal queues. Restrict to the non-reserved nodes.
if job_set <= set([str(node_id) for node_id in
self.nodes_by_queue[job['queue']]]):
node_id_list = requested_locations
if not set(node_id_list).isdisjoint(forbidden):
......@@ -725,11 +727,11 @@ class CraySystem(BaseSystem):
# active reservation. Remove locaitons and drop available
# nodecount appropriately.
node_id_list = list(set(node_id_list) - forbidden)
idle_nodecount = len(node_id_list)
else:
idle_nodecount = 0
node_id_list = []
raise ValueError
if not requested_loc_in_forbidden:
raise ValueError("forbidden locations not in queue")
idle_nodecount = len(node_id_list)
return (idle_nodecount, node_id_list)
@locking
......
......@@ -162,6 +162,16 @@ class TestCraySystem(object):
assert nodecount == 1, 'Wrong nodecount'
assert nodelist == ['3'], 'Wrong node in list %s' % nodelist
@raises(ValueError)
def test_assemble_queue_data_attrs_bad_location(self):
'''_assemble_queue_data: raise error for location completely outside of
queue'''
self.base_job['attrs'] = {'location':'6'}
nodecount, nodelist = self.system._assemble_queue_data(self.base_job,
self.system._idle_nodes_by_queue())
assert nodecount == 1, 'Wrong nodecount'
assert nodelist == ['3'], 'Wrong node in list %s' % nodelist
def test_assemble_queue_data_attrs_location_multi(self):
'''_assemble_queue_data: return only attr locaiton complex loc string'''
self.base_job['attrs'] = {'location':'1-3,5'}
......@@ -236,3 +246,4 @@ class TestCraySystem(object):
self.system._idle_nodes_by_queue())
assert nodecount == 3, 'Wrong nodecount %s' % nodecount
assert sorted(nodelist) == ['1','2','4'], 'Wrong nodes in list %s' % nodelist
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment