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 = 4000 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 guests_uuid_mapping = dict() rows, _ = GuestCPUMemory.get_by_filter(limit=limit, filter_str=filter_str) for row in rows: if row['guest_uuid'] not in guests_uuid_mapping: guests_uuid_mapping[row['guest_uuid']] = { 'cpu_load': 0, 'count': 0 } guests_uuid_mapping[row['guest_uuid']]['cpu_load'] += row['cpu_load'] guests_uuid_mapping[row['guest_uuid']]['count'] += 1.0 rows = list() for k, v in guests_uuid_mapping.items(): # 忽略除数为 0 的情况 if v['cpu_load'] == 0: continue rows.append({'guest_uuid': 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 guests_uuid_mapping.clear() rows, _ = GuestDiskIO.get_by_filter(limit=limit, filter_str=filter_str) for row in rows: if row['disk_uuid'] not in guests_uuid_mapping: guests_uuid_mapping[row['disk_uuid']] = { 'rw_bytes': 0, 'rw_req': 0 } guests_uuid_mapping[ row['disk_uuid']]['rw_bytes'] += row['rd_bytes'] + row['wr_bytes'] guests_uuid_mapping[ row['disk_uuid']]['rw_req'] += row['rd_req'] + row['wr_req'] rows = list() for k, v in guests_uuid_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 }) 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 guests_uuid_mapping.clear() rows, _ = GuestTraffic.get_by_filter(limit=limit, filter_str=filter_str) for row in rows: if row['guest_uuid'] not in guests_uuid_mapping: guests_uuid_mapping[row['guest_uuid']] = { 'rt_bytes': 0, 'rt_packets': 0 } guests_uuid_mapping[ row['guest_uuid']]['rt_bytes'] += row['rx_bytes'] + row['tx_bytes'] guests_uuid_mapping[row['guest_uuid']][ 'rt_packets'] += row['rx_packets'] + row['tx_packets'] rows = list() for k, v in guests_uuid_mapping.items(): # 过滤掉无流量的数据 if v['rt_packets'] == 0: continue rows.append({ 'guest_uuid': k, 'rt_bytes': v['rt_bytes'] * 60 * _range, 'rt_packets': v['rt_packets'] * 60 * _range }) 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 Guests 容量为 4000 个 volume = 4000 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, _ = GuestCPUMemory.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, _ = GuestDiskIO.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, _ = GuestTraffic.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() snapshot = Snapshot() snapshot_disk_mapping = SnapshotDiskMapping() os_template_image = OSTemplateImage() config = Config() config.id = 1 guest_cpu_memory = GuestCPUMemory() guest_traffic = GuestTraffic() guest_disk_io = GuestDiskIO() 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.node_id = cls.message['node_id'] 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: if cls.message['message']['progress'] <= cls.guest.progress: return cls.guest.progress = cls.message['message']['progress'] elif cls.guest.status == GuestState.snapshot_converting.value: cls.os_template_image.id = cls.message['message'][ 'os_template_image_id'] cls.os_template_image.get() if cls.message['message'][ 'progress'] <= cls.os_template_image.progress: return cls.os_template_image.progress = cls.message['message']['progress'] cls.os_template_image.update() return 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({'node_id': cls.guest.node_id}, 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'], 'nonrandom': False, 'threads_status': cls.message['message']['threads_status'], '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'] node_id = cls.message['node_id'] 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.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) SSHKeyGuestMapping.delete_by_filter(filter_str=':'.join( ['guest_uuid', 'eq', cls.guest.uuid])) elif action == 'reset_password': if state == ResponseState.success.value: cls.guest.uuid = uuid cls.guest.get_by('uuid') cls.guest.password = cls.message['message'][ 'passback_parameters']['password'] cls.guest.update() 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': if state == ResponseState.success.value: pass elif _object == 'disk': if action == 'create': cls.disk.uuid = uuid cls.disk.get_by('uuid') cls.disk.node_id = node_id 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.config.get() cls.disk.uuid = uuid cls.disk.get_by('uuid') cls.disk.size = cls.message['message'][ 'passback_parameters']['size'] cls.disk.quota(config=cls.config) cls.disk.update() elif action == 'delete': cls.disk.uuid = uuid cls.disk.get_by('uuid') cls.disk.delete() elif _object == 'snapshot': if action == 'create': cls.snapshot.id = cls.message['message'][ 'passback_parameters']['id'] cls.snapshot.get() if state == ResponseState.success.value: cls.snapshot.snapshot_id = data['snapshot_id'] cls.snapshot.parent_id = data['parent_id'] cls.snapshot.xml = data['xml'] cls.snapshot.progress = 100 cls.snapshot.update() disks, _ = Disk.get_by_filter(filter_str='guest_uuid:eq:' + cls.snapshot.guest_uuid) for disk in disks: cls.snapshot_disk_mapping.snapshot_id = cls.snapshot.snapshot_id cls.snapshot_disk_mapping.disk_uuid = disk['uuid'] cls.snapshot_disk_mapping.create() else: cls.snapshot.progress = 255 cls.snapshot.update() if action == 'delete': if state == ResponseState.success.value: cls.snapshot.id = cls.message['message'][ 'passback_parameters']['id'] cls.snapshot.get() # 更新子快照的 parent_id 为,当前快照的 parent_id。因为当前快照已被删除。 Snapshot.update_by_filter( {'parent_id': cls.snapshot.parent_id}, filter_str='parent_id:eq:' + cls.snapshot.snapshot_id) SnapshotDiskMapping.delete_by_filter(filter_str=':'.join( ['snapshot_id', 'eq', cls.snapshot.snapshot_id])) cls.snapshot.delete() else: pass if action == 'revert': # 不论恢复成功与否,都使快照恢复至正常状态。 cls.snapshot.id = cls.message['message'][ 'passback_parameters']['id'] cls.snapshot.get() cls.snapshot.progress = 100 cls.snapshot.update() if action == 'convert': cls.snapshot.snapshot_id = cls.message['message'][ 'passback_parameters']['id'] cls.snapshot.get_by('snapshot_id') cls.snapshot.progress = 100 cls.snapshot.update() cls.os_template_image.id = cls.message['message'][ 'passback_parameters']['os_template_image_id'] cls.os_template_image.get() if state == ResponseState.success.value: cls.os_template_image.progress = 100 else: cls.os_template_image.progress = 255 cls.os_template_image.update() elif _object == 'os_template_image': if action == 'delete': cls.os_template_image.id = cls.message['message'][ 'passback_parameters']['id'] cls.os_template_image.get() if state == ResponseState.success.value: cls.os_template_image.delete() else: pass else: pass @classmethod def guest_collection_performance_processor(cls): data_kind = cls.message['type'] timestamp = ji.Common.ts() timestamp -= (timestamp % 60) data = cls.message['message']['data'] if data_kind == GuestCollectionPerformanceDataKind.cpu_memory.value: for item in data: cls.guest_cpu_memory.guest_uuid = item['guest_uuid'] cls.guest_cpu_memory.cpu_load = item['cpu_load'] cls.guest_cpu_memory.memory_available = item[ 'memory_available'] cls.guest_cpu_memory.memory_unused = item['memory_unused'] cls.guest_cpu_memory.timestamp = timestamp cls.guest_cpu_memory.create() if data_kind == GuestCollectionPerformanceDataKind.traffic.value: for item in data: cls.guest_traffic.guest_uuid = item['guest_uuid'] cls.guest_traffic.name = item['name'] cls.guest_traffic.rx_bytes = item['rx_bytes'] cls.guest_traffic.rx_packets = item['rx_packets'] cls.guest_traffic.rx_errs = item['rx_errs'] cls.guest_traffic.rx_drop = item['rx_drop'] cls.guest_traffic.tx_bytes = item['tx_bytes'] cls.guest_traffic.tx_packets = item['tx_packets'] cls.guest_traffic.tx_errs = item['tx_errs'] cls.guest_traffic.tx_drop = item['tx_drop'] cls.guest_traffic.timestamp = timestamp cls.guest_traffic.create() if data_kind == GuestCollectionPerformanceDataKind.disk_io.value: for item in data: cls.guest_disk_io.disk_uuid = item['disk_uuid'] cls.guest_disk_io.rd_req = item['rd_req'] cls.guest_disk_io.rd_bytes = item['rd_bytes'] cls.guest_disk_io.wr_req = item['wr_req'] cls.guest_disk_io.wr_bytes = item['wr_bytes'] cls.guest_disk_io.timestamp = timestamp cls.guest_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.guest_collection_performance.value: cls.guest_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()) time.sleep(1)