def clear_expire_monitor_log():
from models import GuestCPUMemory, GuestTraffic, GuestDiskIO, HostCPUMemory, HostTraffic, HostDiskUsageIO
from models import Utils
already_clear = False
the_time = '03:30'
logger.info(msg='Clear expire log monitor is launched.')
while True:
try:
if Utils.exit_flag:
msg = 'Thread clear expire monitor log say bye-bye'
print msg
logger.info(msg=msg)
return
time.sleep(10)
# 每天凌晨3点30分执行，清除15天前的监控记录
if ji.JITime.now_time()[:5] == the_time and not already_clear:
boundary = ji.Common.ts() - 86400 * 15
filter_str = 'timestamp:lt:' + boundary.__str__()
GuestCPUMemory.delete_by_filter(filter_str=filter_str)
GuestTraffic.delete_by_filter(filter_str=filter_str)
GuestDiskIO.delete_by_filter(filter_str=filter_str)
HostCPUMemory.delete_by_filter(filter_str=filter_str)
HostTraffic.delete_by_filter(filter_str=filter_str)
HostDiskUsageIO.delete_by_filter(filter_str=filter_str)
already_clear = True
if already_clear and ji.JITime.now_time()[:5] != the_time:
already_clear = False
except:
logger.error(traceback.format_exc())
def r_last_the_range_minutes_top_10(_range):
volume = 100
limit = volume * _range
length = 10
end_ts = ji.Common.ts() - 60
start_ts = end_ts - 60 * _range
# 避免落在时间边界上，导致过滤条件的范围落空
if start_ts % 60 == 0:
start_ts -= 1
ret = dict()
ret['state'] = ji.Common.exchange_state(20000)
ret['data'] = {
'cpu_load': list(),
'rw_bytes': list(),
'rw_req': list(),
'rt_bytes': list(),
'rt_packets': list()
}
filter_str = ';'.join([
':'.join(['timestamp', 'gt', start_ts.__str__()]),
':'.join(['timestamp', 'lt', end_ts.__str__()])
])
# cpu 负载
hosts_node_id_mapping = dict()
rows, _ = HostCPUMemory.get_by_filter(limit=limit, filter_str=filter_str)
for row in rows:
if row['node_id'] not in hosts_node_id_mapping:
hosts_node_id_mapping[row['node_id']] = {'cpu_load': 0, 'count': 0}
hosts_node_id_mapping[row['node_id']]['cpu_load'] += row['cpu_load']
hosts_node_id_mapping[row['node_id']]['count'] += 1.0
rows = list()
for k, v in hosts_node_id_mapping.items():
# 忽略除数为 0 的情况
if v['cpu_load'] == 0:
continue
rows.append({'node_id': k, 'cpu_load': v['cpu_load'] / v['count']})
effective_range = length
if rows.__len__() < length:
effective_range = rows.__len__()
rows.sort(key=lambda _k: _k['cpu_load'], reverse=True)
ret['data']['cpu_load'] = rows[0:effective_range]
# 磁盘使用统计
hosts_node_id_mapping.clear()
rows, _ = HostDiskUsageIO.get_by_filter(limit=limit, filter_str=filter_str)
for row in rows:
disk_uuid = '_'.join([row['node_id'].__str__(), row['mountpoint']])
if disk_uuid not in hosts_node_id_mapping:
hosts_node_id_mapping[disk_uuid] = {
'rw_bytes': 0,
'rw_req': 0,
'node_id': row['node_id'],
'mountpoint': row['mountpoint']
}
hosts_node_id_mapping[disk_uuid][
'rw_bytes'] += row['rd_bytes'] + row['wr_bytes']
hosts_node_id_mapping[disk_uuid][
'rw_req'] += row['rd_req'] + row['wr_req']
rows = list()
for k, v in hosts_node_id_mapping.items():
# 过滤掉无操作的数据
if v['rw_req'] == 0:
continue
rows.append({
'disk_uuid': k,
'rw_bytes': v['rw_bytes'] * 60 * _range,
'rw_req': v['rw_req'] * 60 * _range,
'node_id': v['node_id'],
'mountpoint': v['mountpoint']
})
effective_range = length
if rows.__len__() < length:
effective_range = rows.__len__()
rows.sort(key=lambda _k: _k['rw_bytes'], reverse=True)
ret['data']['rw_bytes'] = rows[0:effective_range]
rows.sort(key=lambda _k: _k['rw_req'], reverse=True)
ret['data']['rw_req'] = rows[0:effective_range]
# 网络流量
hosts_node_id_mapping.clear()
rows, _ = HostTraffic.get_by_filter(limit=limit, filter_str=filter_str)
for row in rows:
nic_uuid = '_'.join([row['node_id'].__str__(), row['name']])
if nic_uuid not in hosts_node_id_mapping:
hosts_node_id_mapping[nic_uuid] = {
'rt_bytes': 0,
'rt_packets': 0,
'node_id': row['node_id'],
'name': row['name']
}
hosts_node_id_mapping[nic_uuid][
'rt_bytes'] += row['rx_bytes'] + row['tx_bytes']
hosts_node_id_mapping[nic_uuid][
'rt_packets'] += row['rx_packets'] + row['tx_packets']
rows = list()
for k, v in hosts_node_id_mapping.items():
# 过滤掉无流量的数据
if v['rt_packets'] == 0:
continue
rows.append({
'nic_uuid': k,
'rt_bytes': v['rt_bytes'] * 60 * _range,
'rt_packets': v['rt_packets'] * 60 * _range,
'node_id': v['node_id'],
'name': v['name']
})
effective_range = length
if rows.__len__() < length:
effective_range = rows.__len__()
rows.sort(key=lambda _k: _k['rt_bytes'], reverse=True)
ret['data']['rt_bytes'] = rows[0:effective_range]
rows.sort(key=lambda _k: _k['rt_packets'], reverse=True)
ret['data']['rt_packets'] = rows[0:effective_range]
return ret
def r_current_top_10():
# JimV 设计的 Hosts 容量为 100 个
volume = 100
limit = volume
length = 10
end_ts = ji.Common.ts() - 60
start_ts = end_ts - 60
# 避免落在时间边界上，导致过滤条件的范围落空
if start_ts % 60 == 0:
start_ts -= 1
ret = dict()
ret['state'] = ji.Common.exchange_state(20000)
ret['data'] = {
'cpu_load': list(),
'rw_bytes': list(),
'rw_req': list(),
'rt_bytes': list(),
'rt_packets': list()
}
filter_str = ';'.join([
':'.join(['timestamp', 'gt', start_ts.__str__()]),
':'.join(['timestamp', 'lt', end_ts.__str__()])
])
rows, _ = HostCPUMemory.get_by_filter(limit=limit, filter_str=filter_str)
rows.sort(key=lambda k: k['cpu_load'], reverse=True)
effective_range = length
if rows.__len__() < length:
effective_range = rows.__len__()
for i in range(effective_range):
if rows[i]['cpu_load'] == 0:
break
ret['data']['cpu_load'].append(rows[i])
rows, _ = HostDiskUsageIO.get_by_filter(limit=limit, filter_str=filter_str)
for i in range(rows.__len__()):
rows[i]['rw_bytes'] = rows[i]['rd_bytes'] + rows[i]['wr_bytes']
rows[i]['rw_req'] = rows[i]['rd_req'] + rows[i]['wr_req']
effective_range = length
if rows.__len__() < length:
effective_range = rows.__len__()
rows.sort(key=lambda k: k['rw_bytes'], reverse=True)
for i in range(effective_range):
if rows[i]['rw_req'] == 0:
break
ret['data']['rw_bytes'].append(rows[i])
rows.sort(key=lambda k: k['rw_req'], reverse=True)
for i in range(effective_range):
if rows[i]['rw_req'] == 0:
break
ret['data']['rw_req'].append(rows[i])
rows, _ = HostTraffic.get_by_filter(limit=limit, filter_str=filter_str)
for i in range(rows.__len__()):
rows[i]['rt_bytes'] = rows[i]['rx_bytes'] + rows[i]['tx_bytes']
rows[i]['rt_packets'] = rows[i]['rx_packets'] + rows[i]['tx_packets']
effective_range = length
if rows.__len__() < length:
effective_range = rows.__len__()
rows.sort(key=lambda k: k['rt_bytes'], reverse=True)
for i in range(effective_range):
if rows[i]['rt_packets'] == 0:
break
ret['data']['rt_bytes'].append(rows[i])
rows.sort(key=lambda k: k['rt_packets'], reverse=True)
for i in range(effective_range):
if rows[i]['rt_packets'] == 0:
break
ret['data']['rt_packets'].append(rows[i])
return ret
class EventProcessor(object):
message = None
log = Log()
guest = Guest()
guest_migrate_info = GuestMigrateInfo()
disk = Disk()
config = Config()
cpu_memory = CPUMemory()
traffic = Traffic()
disk_io = DiskIO()
host_cpu_memory = HostCPUMemory()
host_traffic = HostTraffic()
host_disk_usage_io = HostDiskUsageIO()
@classmethod
def log_processor(cls):
cls.log.set(type=cls.message['type'],
timestamp=cls.message['timestamp'],
host=cls.message['host'],
message=cls.message['message'],
full_message='' if cls.message['message'].__len__() < 255
else cls.message['message'])
cls.log.create()
@classmethod
def guest_event_processor(cls):
cls.guest.uuid = cls.message['message']['uuid']
cls.guest.get_by('uuid')
cls.guest.on_host = cls.message['host']
last_status = cls.guest.status
cls.guest.status = cls.message['type']
if cls.message['type'] == GuestState.update.value:
# 更新事件不改变 Guest 的状态
cls.guest.status = last_status
cls.guest.xml = cls.message['message']['xml']
elif cls.guest.status == GuestState.migrating.value:
try:
cls.guest_migrate_info.uuid = cls.guest.uuid
cls.guest_migrate_info.get_by('uuid')
cls.guest_migrate_info.type = cls.message['message'][
'migrating_info']['type']
cls.guest_migrate_info.time_elapsed = cls.message['message'][
'migrating_info']['time_elapsed']
cls.guest_migrate_info.time_remaining = cls.message['message'][
'migrating_info']['time_remaining']
cls.guest_migrate_info.data_total = cls.message['message'][
'migrating_info']['data_total']
cls.guest_migrate_info.data_processed = cls.message['message'][
'migrating_info']['data_processed']
cls.guest_migrate_info.data_remaining = cls.message['message'][
'migrating_info']['data_remaining']
cls.guest_migrate_info.mem_total = cls.message['message'][
'migrating_info']['mem_total']
cls.guest_migrate_info.mem_processed = cls.message['message'][
'migrating_info']['mem_processed']
cls.guest_migrate_info.mem_remaining = cls.message['message'][
'migrating_info']['mem_remaining']
cls.guest_migrate_info.file_total = cls.message['message'][
'migrating_info']['file_total']
cls.guest_migrate_info.file_processed = cls.message['message'][
'migrating_info']['file_processed']
cls.guest_migrate_info.file_remaining = cls.message['message'][
'migrating_info']['file_remaining']
cls.guest_migrate_info.update()
except ji.PreviewingError as e:
ret = json.loads(e.message)
if ret['state']['code'] == '404':
cls.guest_migrate_info.type = cls.message['message'][
'migrating_info']['type']
cls.guest_migrate_info.time_elapsed = cls.message[
'message']['migrating_info']['time_elapsed']
cls.guest_migrate_info.time_remaining = cls.message[
'message']['migrating_info']['time_remaining']
cls.guest_migrate_info.data_total = cls.message['message'][
'migrating_info']['data_total']
cls.guest_migrate_info.data_processed = cls.message[
'message']['migrating_info']['data_processed']
cls.guest_migrate_info.data_remaining = cls.message[
'message']['migrating_info']['data_remaining']
cls.guest_migrate_info.mem_total = cls.message['message'][
'migrating_info']['mem_total']
cls.guest_migrate_info.mem_processed = cls.message[
'message']['migrating_info']['mem_processed']
cls.guest_migrate_info.mem_remaining = cls.message[
'message']['migrating_info']['mem_remaining']
cls.guest_migrate_info.file_total = cls.message['message'][
'migrating_info']['file_total']
cls.guest_migrate_info.file_processed = cls.message[
'message']['migrating_info']['file_processed']
cls.guest_migrate_info.file_remaining = cls.message[
'message']['migrating_info']['file_remaining']
cls.guest_migrate_info.create()
elif cls.guest.status == GuestState.creating.value:
cls.guest.progress = cls.message['message']['progress']
cls.guest.update()
# 限定特殊情况下更新磁盘所属 Guest，避免迁移、创建时频繁被无意义的更新
if cls.guest.status in [
GuestState.running.value, GuestState.shutoff.value
]:
cls.disk.update_by_filter({'on_host': cls.guest.on_host},
filter_str='guest_uuid:eq:' +
cls.guest.uuid)
@classmethod
def host_event_processor(cls):
key = cls.message['message']['node_id']
value = {
'hostname': cls.message['host'],
'cpu': cls.message['message']['cpu'],
'system_load': cls.message['message']['system_load'],
'memory': cls.message['message']['memory'],
'memory_available': cls.message['message']['memory_available'],
'interfaces': cls.message['message']['interfaces'],
'disks': cls.message['message']['disks'],
'boot_time': cls.message['message']['boot_time'],
'timestamp': ji.Common.ts()
}
db.r.hset(app.config['hosts_info'],
key=key,
value=json.dumps(value, ensure_ascii=False))
@classmethod
def response_processor(cls):
_object = cls.message['message']['_object']
action = cls.message['message']['action']
uuid = cls.message['message']['uuid']
state = cls.message['type']
data = cls.message['message']['data']
hostname = cls.message['host']
if _object == 'guest':
if action == 'create':
if state == ResponseState.success.value:
# 系统盘的 UUID 与其 Guest 的 UUID 相同
cls.disk.uuid = uuid
cls.disk.get_by('uuid')
cls.disk.guest_uuid = uuid
cls.disk.state = DiskState.mounted.value
# disk_info['virtual-size'] 的单位为Byte，需要除以 1024 的 3 次方，换算成单位为 GB 的值
cls.disk.size = data['disk_info']['virtual-size'] / (1024**
3)
cls.disk.update()
else:
cls.guest.uuid = uuid
cls.guest.get_by('uuid')
cls.guest.status = GuestState.dirty.value
cls.guest.update()
elif action == 'migrate':
pass
elif action == 'delete':
if state == ResponseState.success.value:
cls.config.id = 1
cls.config.get()
cls.guest.uuid = uuid
cls.guest.get_by('uuid')
if IP(cls.config.start_ip).int() <= IP(
cls.guest.ip).int() <= IP(cls.config.end_ip).int():
if db.r.srem(app.config['ip_used_set'], cls.guest.ip):
db.r.sadd(app.config['ip_available_set'],
cls.guest.ip)
if (cls.guest.vnc_port - cls.config.start_vnc_port) <= \
(IP(cls.config.end_ip).int() - IP(cls.config.start_ip).int()):
if db.r.srem(app.config['vnc_port_used_set'],
cls.guest.vnc_port):
db.r.sadd(app.config['vnc_port_available_set'],
cls.guest.vnc_port)
cls.guest.delete()
# TODO: 加入是否删除使用的数据磁盘开关，如果为True，则顺便删除使用的磁盘。否则解除该磁盘被使用的状态。
cls.disk.uuid = uuid
cls.disk.get_by('uuid')
cls.disk.delete()
cls.disk.update_by_filter(
{
'guest_uuid': '',
'sequence': -1,
'state': DiskState.idle.value
},
filter_str='guest_uuid:eq:' + cls.guest.uuid)
elif action == 'attach_disk':
cls.disk.uuid = cls.message['message']['passback_parameters'][
'disk_uuid']
cls.disk.get_by('uuid')
if state == ResponseState.success.value:
cls.disk.guest_uuid = uuid
cls.disk.sequence = cls.message['message'][
'passback_parameters']['sequence']
cls.disk.state = DiskState.mounted.value
cls.disk.update()
elif action == 'detach_disk':
cls.disk.uuid = cls.message['message']['passback_parameters'][
'disk_uuid']
cls.disk.get_by('uuid')
if state == ResponseState.success.value:
cls.disk.guest_uuid = ''
cls.disk.sequence = -1
cls.disk.state = DiskState.idle.value
cls.disk.update()
elif action == 'boot':
boot_jobs_id = cls.message['message']['passback_parameters'][
'boot_jobs_id']
if state == ResponseState.success.value:
cls.guest.uuid = uuid
cls.guest.delete_boot_jobs(boot_jobs_id=boot_jobs_id)
elif _object == 'disk':
if action == 'create':
cls.disk.uuid = uuid
cls.disk.get_by('uuid')
cls.disk.on_host = hostname
if state == ResponseState.success.value:
cls.disk.state = DiskState.idle.value
else:
cls.disk.state = DiskState.dirty.value
cls.disk.update()
elif action == 'resize':
if state == ResponseState.success.value:
cls.disk.uuid = uuid
cls.disk.get_by('uuid')
cls.disk.size = cls.message['message'][
'passback_parameters']['size']
cls.disk.update()
elif action == 'delete':
cls.disk.uuid = uuid
cls.disk.get_by('uuid')
cls.disk.delete()
else:
pass
@classmethod
def collection_performance_processor(cls):
data_kind = cls.message['type']
timestamp = ji.Common.ts()
timestamp -= (timestamp % 60)
data = cls.message['message']['data']
if data_kind == CollectionPerformanceDataKind.cpu_memory.value:
for item in data:
cls.cpu_memory.guest_uuid = item['guest_uuid']
cls.cpu_memory.cpu_load = item['cpu_load']
cls.cpu_memory.memory_available = item['memory_available']
cls.cpu_memory.memory_unused = item['memory_unused']
cls.cpu_memory.timestamp = timestamp
cls.cpu_memory.create()
if data_kind == CollectionPerformanceDataKind.traffic.value:
for item in data:
cls.traffic.guest_uuid = item['guest_uuid']
cls.traffic.name = item['name']
cls.traffic.rx_bytes = item['rx_bytes']
cls.traffic.rx_packets = item['rx_packets']
cls.traffic.rx_errs = item['rx_errs']
cls.traffic.rx_drop = item['rx_drop']
cls.traffic.tx_bytes = item['tx_bytes']
cls.traffic.tx_packets = item['tx_packets']
cls.traffic.tx_errs = item['tx_errs']
cls.traffic.tx_drop = item['tx_drop']
cls.traffic.timestamp = timestamp
cls.traffic.create()
if data_kind == CollectionPerformanceDataKind.disk_io.value:
for item in data:
cls.disk_io.disk_uuid = item['disk_uuid']
cls.disk_io.rd_req = item['rd_req']
cls.disk_io.rd_bytes = item['rd_bytes']
cls.disk_io.wr_req = item['wr_req']
cls.disk_io.wr_bytes = item['wr_bytes']
cls.disk_io.timestamp = timestamp
cls.disk_io.create()
else:
pass
@classmethod
def host_collection_performance_processor(cls):
data_kind = cls.message['type']
timestamp = ji.Common.ts()
timestamp -= (timestamp % 60)
data = cls.message['message']['data']
if data_kind == HostCollectionPerformanceDataKind.cpu_memory.value:
cls.host_cpu_memory.node_id = data['node_id']
cls.host_cpu_memory.cpu_load = data['cpu_load']
cls.host_cpu_memory.memory_available = data['memory_available']
cls.host_cpu_memory.timestamp = timestamp
cls.host_cpu_memory.create()
if data_kind == HostCollectionPerformanceDataKind.traffic.value:
for item in data:
cls.host_traffic.node_id = item['node_id']
cls.host_traffic.name = item['name']
cls.host_traffic.rx_bytes = item['rx_bytes']
cls.host_traffic.rx_packets = item['rx_packets']
cls.host_traffic.rx_errs = item['rx_errs']
cls.host_traffic.rx_drop = item['rx_drop']
cls.host_traffic.tx_bytes = item['tx_bytes']
cls.host_traffic.tx_packets = item['tx_packets']
cls.host_traffic.tx_errs = item['tx_errs']
cls.host_traffic.tx_drop = item['tx_drop']
cls.host_traffic.timestamp = timestamp
cls.host_traffic.create()
if data_kind == HostCollectionPerformanceDataKind.disk_usage_io.value:
for item in data:
cls.host_disk_usage_io.node_id = item['node_id']
cls.host_disk_usage_io.mountpoint = item['mountpoint']
cls.host_disk_usage_io.used = item['used']
cls.host_disk_usage_io.rd_req = item['rd_req']
cls.host_disk_usage_io.rd_bytes = item['rd_bytes']
cls.host_disk_usage_io.wr_req = item['wr_req']
cls.host_disk_usage_io.wr_bytes = item['wr_bytes']
cls.host_disk_usage_io.timestamp = timestamp
cls.host_disk_usage_io.create()
else:
pass
@classmethod
def launch(cls):
logger.info(msg='Thread EventProcessor is launched.')
while True:
if Utils.exit_flag:
msg = 'Thread EventProcessor say bye-bye'
print msg
logger.info(msg=msg)
return
try:
report = db.r.lpop(app.config['upstream_queue'])
if report is None:
time.sleep(1)
continue
cls.message = json.loads(report)
if cls.message['kind'] == EmitKind.log.value:
cls.log_processor()
elif cls.message['kind'] == EmitKind.guest_event.value:
cls.guest_event_processor()
elif cls.message['kind'] == EmitKind.host_event.value:
cls.host_event_processor()
elif cls.message['kind'] == EmitKind.response.value:
cls.response_processor()
elif cls.message[
'kind'] == EmitKind.collection_performance.value:
cls.collection_performance_processor()
elif cls.message[
'kind'] == EmitKind.host_collection_performance.value:
cls.host_collection_performance_processor()
else:
pass
except Exception as e:
logger.error(traceback.format_exc())