stackforge/libra
Code Issues Proposed changes

Fixes to make statsd actually work

Browse Source 
Change-Id: I10f46507ce452e9a9c8352019c008d3698a341d7
This commit is contained in:
Andrew Hutchings 2013-04-18 17:50:08 +01:00
parent 1206a22c5c
commit 184fe83639
3 changed files with 16 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
libra/statsd/admin_api.py
View File

@ -18,6 +18,10 @@ import sys
import json
class APIError(Exception):
pass
class AdminAPI(object):
def __init__(self, addresses, logger):
self.logger = logger
@ -42,7 +46,9 @@ class AdminAPI(object):
limit = 20
lb_list = []
while True:
nodes = self._get_node_list(limit, marker)
success, nodes = self._get_node_list(limit, marker)
if not success:
raise APIError
# if we hit an empty device list we have hit end of list
if not len(nodes['devices']):
break

10
libra/statsd/gearman.py
View File

@ -12,8 +12,6 @@
# License for the specific language governing permissions and limitations
# under the License.
import gearman
from libra.common.json_gearman import JSONGearmanClient
@ -27,23 +25,23 @@ class GearJobs(object):
failed_list = []
job_data = {"hpcs_action": "STATS"}
for node in node_list:
list_of_jobs.append(dict(task=node, data=job_data))
list_of_jobs.append(dict(task=str(node), data=job_data))
submitted_pings = self.gm_client.submit_multiple_jobs(
list_of_jobs, background=False, wait_until_complete=True,
poll_timeout=5.0
)
for ping in submitted_pings:
if ping.state == gearman.JOB_UNKNOWN:
if ping.state == 'UNKNOWN':
# TODO: Gearman server failed, ignoring for now
self.logger.error('Gearman Job server fail')
continue
if ping.timed_out:
# Ping timeout
failed_list.append(ping['task'])
failed_list.append(ping.job.task)
continue
if ping.result['hpcs_response'] == 'FAIL':
# Error returned by Gearman
failed_list.append(ping['task'])
failed_list.append(ping.job.task)
continue
return failed_list

9
libra/statsd/scheduler.py
View File

@ -82,7 +82,7 @@ class Sched(object):
failed_nodes = gearman.send_pings(node_list)
failed = len(failed_nodes)
if failed > 0:
self._send_fails(lb_list, failed_nodes, lb_list)
self._send_fails(failed_nodes, lb_list)
else:
self.logger.error('No working API server found')
return (0, 0)
@ -95,14 +95,15 @@ class Sched(object):
message = (
'Load balancer failed\n'
'ID: {0}\n'
'IP: (1}\n'
'IP: {1}\n'
'tenant: {2}\n'.format(
data['id'], data['floatingIPAddr'],
data['id'], data['floatingIpAddr'],
data['loadBalancers'][0]['hpcs_tenantid']
)
)
for driver in self.drivers:
driver.send_alert(message, data['id'])
instance = driver(self.logger)
instance.send_alert(message, data['id'])
def _get_node(self, node, node_list):
for n in node_list: