def rescheduler_taskgroup(self,
request,
project_id,
cluster_id,
ns,
instance_name,
taskgroup_name,
kind=2):
"""重启更新
"""
if kind == 2:
client = MesosClient(request.user.token.access_token, project_id,
cluster_id, None)
resp = client.rescheduler_mesos_taskgroup(ns, instance_name,
taskgroup_name)
else:
client = K8SClient(request.user.token.access_token, project_id,
cluster_id, None)
resp = client.delete_pod(ns, taskgroup_name)
if resp.get("code") != ErrorCode.NoError:
return APIResponse({
"code": resp.get("code", DEFAULT_ERROR_CODE),
"message": resp.get("message", _("请求出现异常!"))
})
return APIResponse({"message": _("重新调度成功!")})
class MesosDriver:
def __init__(self, request, project_id, cluster_id):
self.request = request
self.project_id = project_id
self.cluster_id = cluster_id
self.client = MesosClient(self.request.user.token.access_token,
self.project_id, self.cluster_id, None)
def host_container_map(self, resp):
host_container_map = {}
for info in resp.get('data') or []:
if info["namespace"] in MESOS_SKIP_NS_LIST:
continue
host_ip = info.get('data', {}).get('hostIP')
container_count = len(info['data']['containerStatuses'])
if host_ip in host_container_map:
host_container_map[host_ip] += container_count
else:
host_container_map[host_ip] = container_count
return host_container_map
def get_unit_info(self, inner_ip, fields, raise_exception=True):
"""get the resource unit info"""
resp = self.client.get_taskgroup(inner_ip, fields=fields)
if resp.get('code') != ErrorCode.NoError:
logger.error("request taskgroup api error, %s",
resp.get("message"))
if raise_exception:
raise error_codes.APIError(resp.get('message'))
return resp
def get_host_container_count(self, host_ips):
field_list = [
'data.containerStatuses.containerID', 'data.hostIP', "namespace"
]
resp = self.get_unit_info(host_ips, ','.join(field_list))
# compose the host container data
return self.host_container_map(resp)
def flatten_container_info(self, inner_ip):
"""flatten container info by inner_ip"""
def iter_container(tg):
for g in tg:
if g.get("namespace") in MESOS_SKIP_NS_LIST:
continue
for d in g['data']['containerStatuses']:
c = {
'name': d['name'],
'image': d['image'],
'status': d['status'].lower(),
'container_id': d['containerID'],
}
yield c
taskgroups = self.get_unit_info(
inner_ip, fields='data,namespace').get('data') or []
containers = sorted(
[i for i in iter_container(taskgroups)],
key=lambda x: constants.DockerStatusOrdering.get(
x['status'], constants.DockerStatusDefaultOrder),
)
return containers
def disable_node(self, ip):
node_resp = self.client.disable_agent(ip)
if node_resp.get('code') != ErrorCode.NoError:
raise error_codes.APIError(node_resp.get('message'))
def enable_node(self, ip):
node_resp = self.client.enable_agent(ip)
if node_resp.get('code') != ErrorCode.NoError:
raise error_codes.APIError(node_resp.get('message'))
def get_host_unit_list(self, ip, raise_exception=True):
"""get exist pods on the node"""
unit_list = []
fields = 'namespace,resourceName,data.rcname'
resp = self.get_unit_info([ip],
fields,
raise_exception=raise_exception)
for i in resp.get('data') or []:
unit_list.append({
'namespace': i.get('namespace'),
'app_name': i.get('data', {}).get('rcname'),
'taskgroup_name': i.get('resourceName'),
})
return unit_list
def reschedule_pod(self, pod_info, raise_exception=True):
resp = self.client.rescheduler_mesos_taskgroup(
pod_info['namespace'], pod_info['app_name'],
pod_info['taskgroup_name'])
if resp.get('code') != ErrorCode.NoError:
logger.error("request rescheduler taskgroup api error, %s",
resp.get("message"))
if raise_exception:
raise error_codes.APIError(resp.get('message'))
return resp
def reschedule_host_pods(self, ip, raise_exception=True):
unit_list = self.get_host_unit_list(ip,
raise_exception=raise_exception)
for info in unit_list:
self.reschedule_pod(info, raise_exception=raise_exception)
return
