From d910c3f3c4331c34b11eb9982f2955881411e0ae Mon Sep 17 00:00:00 2001 From: iSecloud <869820505@qq.com> Date: Wed, 20 Nov 2024 20:18:52 +0800 Subject: [PATCH] =?UTF-8?q?fix(backend):=20=E5=8D=95=E6=8D=AE=E6=A0=87?= =?UTF-8?q?=E5=87=86=E5=8C=96=E5=8D=8F=E8=AE=AE=E4=BF=AE=E6=94=B9=20#7747?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- dbm-ui/backend/db_dirty/handlers.py | 182 +----------------- dbm-ui/backend/db_dirty/models.py | 21 +- dbm-ui/backend/db_meta/api/machine/apis.py | 6 +- ..._20241014_1042_0043_auto_20241015_2128.py} | 2 +- dbm-ui/backend/db_meta/models/cluster.py | 10 +- dbm-ui/backend/db_package/views.py | 6 +- .../mysql_migrate_cluster_remote_flow.py | 2 +- dbm-ui/backend/ticket/builders/__init__.py | 13 +- dbm-ui/backend/ticket/builders/common/base.py | 4 +- .../builders/mysql/mysql_fixpoint_rollback.py | 15 +- .../builders/mysql/mysql_migrate_cluster.py | 32 +-- .../builders/mysql/mysql_migrate_upgrade.py | 90 ++++++--- dbm-ui/backend/ticket/builders/redis/base.py | 17 +- .../builders/redis/redis_toolbox_cut_off.py | 1 - .../redis/redis_toolbox_proxy_scale_down.py | 32 +-- .../redis/redis_toolbox_shard_update.py | 3 + .../redis/redis_toolbox_type_update.py | 5 +- dbm-ui/backend/ticket/models/ticket.py | 2 + 18 files changed, 178 insertions(+), 265 deletions(-) rename dbm-ui/backend/db_meta/migrations/{0044_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py => 0045_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py} (82%) diff --git a/dbm-ui/backend/db_dirty/handlers.py b/dbm-ui/backend/db_dirty/handlers.py index fc2004179a..2fe7c00d20 100644 --- a/dbm-ui/backend/db_dirty/handlers.py +++ b/dbm-ui/backend/db_dirty/handlers.py @@ -10,28 +10,18 @@ """ import itertools import logging -from collections import defaultdict -from typing import Any, Dict, List +from typing import List from django.utils.translation import ugettext as _ -from backend import env -from backend.components import CCApi -from backend.configuration.constants import SystemSettingsEnum -from backend.configuration.models import SystemSettings from backend.db_dirty.constants import MachineEventType, PoolType from backend.db_dirty.exceptions import PoolTransferException from backend.db_dirty.models import DirtyMachine, MachineEvent -from backend.db_meta.models import AppCache -from backend.db_services.ipchooser.constants import IDLE_HOST_MODULE -from backend.db_services.ipchooser.handlers.topo_handler import TopoHandler -from backend.db_services.ipchooser.query.resource import ResourceQueryHelper from backend.flow.consts import FAILED_STATES from backend.flow.utils.cc_manage import CcManage from backend.ticket.builders import BuilderFactory from backend.ticket.builders.common.base import fetch_apply_hosts from backend.ticket.models import Flow, Ticket -from backend.utils.batch_request import request_multi_thread logger = logging.getLogger("root") @@ -67,172 +57,14 @@ def transfer_hosts_to_pool(cls, operator: str, bk_host_ids: List[int], source: P raise PoolTransferException(_("{}--->{}转移不合法").format(source, target)) @classmethod - def query_dirty_machine_records(cls, bk_host_ids: List[int]): - """ - 查询污点池主机信息 TODO: 污点池废弃，代码将被移除 - @param bk_host_ids: 主机列表 - """ - - def get_module_data(data): - params, res = data - params = params["params"] - return [{"bk_biz_id": params["bk_biz_id"], **d} for d in res] - - if not bk_host_ids: - return [] - - # 如果传入的列表已经是DirtyMachine，则直接用 - if not isinstance(bk_host_ids[0], DirtyMachine): - dirty_machines = DirtyMachine.objects.filter(bk_host_id__in=bk_host_ids) - else: - dirty_machines = bk_host_ids - bk_host_ids = [dirty.bk_host_id for dirty in dirty_machines] - - # 缓存云区域和业务信息 - bk_biz_ids = [dirty_machine.bk_biz_id for dirty_machine in dirty_machines] - for_biz_infos = AppCache.batch_get_app_attr(bk_biz_ids=bk_biz_ids, attr_name="bk_biz_name") - cloud_info = ResourceQueryHelper.search_cc_cloud(get_cache=True) - - # 查询污点主机当前所处的模块 - host_topo_infos = CCApi.find_host_biz_relations(params={"bk_host_id": bk_host_ids}) - host__topo_info_map: Dict[int, List] = defaultdict(list) - biz__modules_map: Dict[int, List] = defaultdict(list) - for topo in host_topo_infos: - host__topo_info_map[topo["bk_host_id"]].append(topo) - biz__modules_map[topo["bk_biz_id"]].append(topo["bk_module_id"]) - # 批量获取业务下模块信息 - module_infos = request_multi_thread( - func=CCApi.find_module_batch, - params_list=[ - { - "params": {"bk_biz_id": biz, "bk_ids": modules, "fields": ["bk_module_id", "bk_module_name"]}, - "use_admin": True, - } - for biz, modules in biz__modules_map.items() - ], - get_data=get_module_data, - in_order=True, - ) - module_infos = list(itertools.chain(*module_infos)) - biz__module__module_name: Dict[int, Dict[int, str]] = defaultdict(dict) - for info in module_infos: - biz__module__module_name[info["bk_biz_id"]][info["bk_module_id"]] = info["bk_module_name"] - - # 获取污点池模块 - system_manage_topo = SystemSettings.get_setting_value(key=SystemSettingsEnum.MANAGE_TOPO.value) - dirty_module = system_manage_topo["dirty_module_id"] - - # 获取污点池列表信息 - dirty_machine_list: List[Dict] = [] - for dirty in dirty_machines: - # 填充污点池主机基础信息 - dirty_machine_info = { - "ip": dirty.ip, - "bk_host_id": dirty.bk_host_id, - "bk_cloud_name": cloud_info[str(dirty.bk_cloud_id)]["bk_cloud_name"], - "bk_cloud_id": dirty.bk_cloud_id, - "bk_biz_name": for_biz_infos[int(dirty.bk_biz_id)], - "bk_biz_id": dirty.bk_biz_id, - "ticket_type": dirty.ticket.ticket_type, - "ticket_id": dirty.ticket.id, - "ticket_type_display": dirty.ticket.get_ticket_type_display(), - "task_id": dirty.flow.flow_obj_id, - "operator": dirty.ticket.creator, - "is_dirty": True, - } - - # 如果主机已经不存在于cc，则仅能删除记录 - if dirty.bk_host_id not in host__topo_info_map: - dirty_machine_info.update(is_dirty=False) - dirty_machine_list.append(dirty_machine_info) - continue - - # 补充主机所在的模块信息 - host_in_module = [ - { - "bk_module_id": h["bk_module_id"], - "bk_module_name": biz__module__module_name[h["bk_biz_id"]].get(h["bk_module_id"], ""), - } - for h in host__topo_info_map[dirty.bk_host_id] - ] - dirty_machine_info.update(bk_module_infos=host_in_module) - - # 如果主机 不处于/不仅仅处于【污点池】中，则不允许移入待回收 - host = host__topo_info_map[dirty.bk_host_id][0] - if len(host__topo_info_map[dirty.bk_host_id]) > 1: - dirty_machine_info.update(is_dirty=False) - elif host["bk_biz_id"] != env.DBA_APP_BK_BIZ_ID or host["bk_module_id"] != dirty_module: - dirty_machine_info.update(is_dirty=False) - - dirty_machine_list.append(dirty_machine_info) - - dirty_machine_list.sort(key=lambda x: x["ticket_id"], reverse=True) - return dirty_machine_list - - @classmethod - def insert_dirty_machines(cls, bk_biz_id: int, bk_host_ids: List[Dict[str, Any]], ticket: Ticket, flow: Flow): + def handle_dirty_machine(cls, ticket_id, root_id, origin_tree_status, target_tree_status): """ - 将机器导入到污点池中 TODO: 污点池废弃，代码将被移除 - @param bk_biz_id: 业务ID - @param bk_host_ids: 主机列表 - @param ticket: 关联的单据 - @param flow: 关联的flow任务 + 处理执行失败/重试成功涉及的污点池机器 + @param ticket_id: 单据ID + @param root_id: 流程ID + @param origin_tree_status: 流程源状态 + @param target_tree_status: 流程目标状态 """ - # 查询污点机器信息 - host_property_filter = { - "condition": "AND", - "rules": [{"field": "bk_host_id", "operator": "in", "value": bk_host_ids}], - } - dirty_host_infos = CCApi.list_hosts_without_biz( - { - # 默认一次性录入的机器不会超过500 - "page": {"start": 0, "limit": 500, "sort": "bk_host_id"}, - "host_property_filter": host_property_filter, - "fields": ["bk_host_id", "bk_cloud_id", "bk_host_innerip"], - }, - use_admin=True, - )["info"] - - # 获取业务空闲机模块，资源池模块和污点池模块 - idle_module = CcManage(bk_biz_id, "").get_biz_internal_module(bk_biz_id)[IDLE_HOST_MODULE]["bk_module_id"] - system_manage_topo = SystemSettings.get_setting_value(key=SystemSettingsEnum.MANAGE_TOPO.value) - resource_module, dirty_module = system_manage_topo["resource_module_id"], system_manage_topo["dirty_module_id"] - # 获取主机的拓扑信息(注：这里不能带上业务信息，因为主机可能转移业务) - host_topo_infos = TopoHandler.query_host_set_module(bk_host_ids=bk_host_ids)["hosts_topo_info"] - # 将污点机器信息转移至DBA污点池模(如果污点机器不在空闲机/资源池，则放弃转移，认为已到正确拓扑) - transfer_host_ids = [ - info["bk_host_id"] - for info in host_topo_infos - if not set(info["bk_module_ids"]) - {resource_module, idle_module} - ] - if transfer_host_ids: - update_host_properties = {"dbm_meta": [], "need_monitor": False, "update_operator": False} - CcManage(bk_biz_id=env.DBA_APP_BK_BIZ_ID, cluster_type="").transfer_host_module( - transfer_host_ids, target_module_ids=[dirty_module], update_host_properties=update_host_properties - ) - - # 录入污点池表中 - exist_dirty_machine_ids = list( - DirtyMachine.objects.filter(bk_host_id__in=bk_host_ids).values_list("bk_host_id", flat=True) - ) - DirtyMachine.objects.bulk_create( - [ - DirtyMachine( - ticket=ticket, - flow=flow, - ip=host["bk_host_innerip"], - bk_biz_id=bk_biz_id, - bk_host_id=host["bk_host_id"], - bk_cloud_id=host["bk_cloud_id"], - ) - for host in dirty_host_infos - if host["bk_host_id"] not in exist_dirty_machine_ids - ] - ) - - @classmethod - def handle_dirty_machine(cls, ticket_id, root_id, origin_tree_status, target_tree_status): - """处理执行失败/重试成功涉及的污点池机器""" if (origin_tree_status not in FAILED_STATES) and (target_tree_status not in FAILED_STATES): return diff --git a/dbm-ui/backend/db_dirty/models.py b/dbm-ui/backend/db_dirty/models.py index f083086e6e..aded49753e 100644 --- a/dbm-ui/backend/db_dirty/models.py +++ b/dbm-ui/backend/db_dirty/models.py @@ -59,15 +59,18 @@ def hosts_pool_transfer(cls, bk_biz_id, hosts, pool, operator="", ticket=None): host_ids = [host["bk_host_id"] for host in hosts] # 主机转入污点/故障池，说明第一次被纳管到池 - if pool in [PoolType.Fault, PoolType.Dirty]: - hosts_pool = [ - cls(bk_biz_id=bk_biz_id, pool=pool, ticket=ticket, creator=operator, updater=operator, **host) - for host in hosts - ] - cls.objects.bulk_create(hosts_pool) - # 待回收只会从故障池转移 - elif pool == PoolType.Recycle: - cls.objects.filter(bk_host_id__in=host_ids).update(pool=pool, ticket=ticket) + # 待回收会从故障池、资源池转移 + # 因此这里判断主机不存在就创建，否则更新 + if pool in [PoolType.Fault, PoolType.Dirty, PoolType.Recycle]: + handle_hosts = cls.objects.filter(bk_host_id__in=host_ids) + if handle_hosts.count() == len(host_ids): + handle_hosts.update(pool=pool, ticket=ticket) + else: + handle_hosts = [ + cls(bk_biz_id=bk_biz_id, pool=pool, ticket=ticket, creator=operator, updater=operator, **host) + for host in hosts + ] + cls.objects.bulk_create(handle_hosts) # 回收机器只能从待回收转移，删除池纳管记录 # 重导入回资源池，删除池纳管记录 elif pool in [PoolType.Recycled, PoolType.Resource]: diff --git a/dbm-ui/backend/db_meta/api/machine/apis.py b/dbm-ui/backend/db_meta/api/machine/apis.py index 7412c8717b..7737fe31da 100644 --- a/dbm-ui/backend/db_meta/api/machine/apis.py +++ b/dbm-ui/backend/db_meta/api/machine/apis.py @@ -193,14 +193,16 @@ def clear_info_for_machine(machines: Optional[List]): # 清理proxy相关信息 for p in proxys: - p.tendbclusterspiderext.delete() + if hasattr(p, "tendbclusterspiderext"): + p.tendbclusterspiderext.delete() p.delete(keep_parents=True) # 清理storage相关信息 for s in storages: for info in StorageInstanceTuple.objects.filter(Q(ejector=s) | Q(receiver=s)): # 先删除额外关联信息，否则会报ProtectedError 异常 - info.tendbclusterstorageset.delete() + if hasattr(info, "tendbclusterstorageset"): + info.tendbclusterstorageset.delete() info.delete() s.delete(keep_parents=True) machine.delete(keep_parents=True) diff --git a/dbm-ui/backend/db_meta/migrations/0044_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py b/dbm-ui/backend/db_meta/migrations/0045_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py similarity index 82% rename from dbm-ui/backend/db_meta/migrations/0044_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py rename to dbm-ui/backend/db_meta/migrations/0045_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py index c6fe2d2ee4..b341e2d078 100644 --- a/dbm-ui/backend/db_meta/migrations/0044_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py +++ b/dbm-ui/backend/db_meta/migrations/0045_merge_0043_auto_20241014_1042_0043_auto_20241015_2128.py @@ -6,7 +6,7 @@ class Migration(migrations.Migration): dependencies = [ - ("db_meta", "0043_auto_20241014_1042"), + ("db_meta", "0044_deviceclass"), ("db_meta", "0043_auto_20241015_2128"), ] diff --git a/dbm-ui/backend/db_meta/models/cluster.py b/dbm-ui/backend/db_meta/models/cluster.py index 03c18c4c58..20c3ec744c 100644 --- a/dbm-ui/backend/db_meta/models/cluster.py +++ b/dbm-ui/backend/db_meta/models/cluster.py @@ -402,16 +402,18 @@ def enable_dbha(self): self.refresh_from_db() @classmethod - def get_cluster_related_machines(cls, cluster_ids: List[int]) -> List: + def get_cluster_related_machines(cls, cluster_ids: List[int], role: str = None) -> List: """ 通过集群id查询集群关联的所有主机信息，即实例所在的主机 """ from backend.db_meta.models import Machine clusters = Cluster.objects.filter(id__in=cluster_ids) - host_ids = set(clusters.values_list("storageinstance__machine__bk_host_id", flat=True)) | set( - clusters.values_list("proxyinstance__machine__bk_host_id", flat=True) - ) + host_ids = set() + if not role or role == "backend": + host_ids |= set(clusters.values_list("storageinstance__machine__bk_host_id", flat=True)) + if not role or role == "proxy": + host_ids |= set(clusters.values_list("proxyinstance__machine__bk_host_id", flat=True)) machines = Machine.objects.filter(bk_host_id__in=host_ids) return machines diff --git a/dbm-ui/backend/db_package/views.py b/dbm-ui/backend/db_package/views.py index 9bf4e6be2e..61d7688a34 100644 --- a/dbm-ui/backend/db_package/views.py +++ b/dbm-ui/backend/db_package/views.py @@ -33,6 +33,7 @@ SyncMediumSerializer, UploadPackageSerializer, ) +from backend.exceptions import ApiRequestError from backend.flow.consts import MediumEnum from backend.iam_app.dataclass import ResourceEnum from backend.iam_app.dataclass.actions import ActionEnum @@ -168,7 +169,10 @@ def partial_update(self, request, *args, **kwargs): ) def destroy(self, request, *args, **kwargs): # 删除制品库文件 - StorageHandler().delete_file(self.get_object().path) + try: + StorageHandler().delete_file(self.get_object().path) + except ApiRequestError as e: + logger.error(_("文件删除异常，错误信息: {}").format(e)) # 删除本地记录 super().destroy(request, *args, **kwargs) return Response() diff --git a/dbm-ui/backend/flow/engine/bamboo/scene/mysql/mysql_migrate_cluster_remote_flow.py b/dbm-ui/backend/flow/engine/bamboo/scene/mysql/mysql_migrate_cluster_remote_flow.py index b981566c42..7d21403d1f 100644 --- a/dbm-ui/backend/flow/engine/bamboo/scene/mysql/mysql_migrate_cluster_remote_flow.py +++ b/dbm-ui/backend/flow/engine/bamboo/scene/mysql/mysql_migrate_cluster_remote_flow.py @@ -234,7 +234,7 @@ def migrate_cluster_flow(self, use_for_upgrade=False): logger.error("cluster {} backup info not exists".format(cluster_model.id)) raise TendbGetBackupInfoFailedException(message=_("获取集群 {} 的备份信息失败".format(cluster_id))) cluster["backupinfo"] = backup_info - cluster["new_master_ip"] = self.data["new_master_ ip"] + cluster["new_master_ip"] = self.data["new_master_ip"] cluster["new_slave_ip"] = self.data["new_slave_ip"] cluster["new_master_port"] = master_model.port cluster["new_slave_port"] = master_model.port diff --git a/dbm-ui/backend/ticket/builders/__init__.py b/dbm-ui/backend/ticket/builders/__init__.py index 58badc4a43..e39cee837d 100644 --- a/dbm-ui/backend/ticket/builders/__init__.py +++ b/dbm-ui/backend/ticket/builders/__init__.py @@ -255,7 +255,9 @@ class RecycleParamBuilder(FlowParamBuilder): controller_map = { DBType.MySQL.value: "MySQLController.mysql_machine_clear_scene", - DBType.TenDBCluster.value: "MySQLController.mysql_machine_clear_scene", + DBType.TenDBCluster.value: "SpiderController.tendbcluster_machine_clear_scene", + # TODO redis清理流程暂时没有 + DBType.Redis.value: "", } def __init__(self, ticket: Ticket): @@ -264,10 +266,15 @@ def __init__(self, ticket: Ticket): assert self.ip_dest is not None def build_controller_info(self) -> dict: + # TODO: 暂时兼容没有清理流程的组件，默认用mysql db_type = self.ticket_data["db_type"] - class_name, flow_name = self.controller_map[db_type].split(".") - module = importlib.import_module(f"backend.flow.engine.controller.{db_type}") + clear_db_type = db_type if self.controller_map.get(db_type) else DBType.MySQL.value + + class_name, flow_name = self.controller_map[clear_db_type].split(".") + module_file_name = "spider" if clear_db_type == DBType.TenDBCluster else clear_db_type + module = importlib.import_module(f"backend.flow.engine.controller.{module_file_name}") self.controller = getattr(getattr(module, class_name), flow_name) + return super().build_controller_info() def format_ticket_data(self): diff --git a/dbm-ui/backend/ticket/builders/common/base.py b/dbm-ui/backend/ticket/builders/common/base.py index c5d21d3513..66e497dc84 100644 --- a/dbm-ui/backend/ticket/builders/common/base.py +++ b/dbm-ui/backend/ticket/builders/common/base.py @@ -505,10 +505,10 @@ def patch_recycle_host_details(self): return self.ticket.details["recycle_hosts"] = ResourceHandler.standardized_resource_host(recycle_hosts, bk_biz_id) - def patch_recycle_cluster_details(self): + def patch_recycle_cluster_details(self, role=None): """补充集群下架后回收主机信息，在下架类单据一定调用此方法""" bk_biz_id = self.ticket.bk_biz_id - recycle_hosts = Cluster.get_cluster_related_machines(fetch_cluster_ids(self.ticket.details)) + recycle_hosts = Cluster.get_cluster_related_machines(fetch_cluster_ids(self.ticket.details), role) recycle_hosts = [{"bk_host_id": host.bk_host_id} for host in recycle_hosts] self.ticket.details["recycle_hosts"] = ResourceHandler.standardized_resource_host(recycle_hosts, bk_biz_id) diff --git a/dbm-ui/backend/ticket/builders/mysql/mysql_fixpoint_rollback.py b/dbm-ui/backend/ticket/builders/mysql/mysql_fixpoint_rollback.py index 0475089988..e0de4c6227 100644 --- a/dbm-ui/backend/ticket/builders/mysql/mysql_fixpoint_rollback.py +++ b/dbm-ui/backend/ticket/builders/mysql/mysql_fixpoint_rollback.py @@ -57,7 +57,9 @@ class FixPointRollbackSerializer(serializers.Serializer): @classmethod def validate_rollback_info(cls, rollback_cluster_type, info, now): # 校验回档集群类型参数 - if rollback_cluster_type == RollbackBuildClusterType.BUILD_INTO_NEW_CLUSTER and not info.get("rollback_host"): + if rollback_cluster_type == RollbackBuildClusterType.BUILD_INTO_NEW_CLUSTER and not ( + info.get("rollback_host") or info.get("resource_spec") + ): raise serializers.ValidationError(_("请提供部署新集群的机器信息")) if rollback_cluster_type != RollbackBuildClusterType.BUILD_INTO_NEW_CLUSTER and not info.get( @@ -102,8 +104,9 @@ def format_ticket_data(self): info["rollback_type"] = f"{info['backup_source'].upper()}_AND_{op_type}" # 格式化定点回档部署的信息 if rollback_cluster_type == RollbackBuildClusterType.BUILD_INTO_NEW_CLUSTER: - info["rollback_ip"] = info["rollback_host"]["ip"] - info["bk_rollback"] = info.pop("rollback_host") + if self.ticket_data["ip_source"] == IpSource.MANUAL_INPUT: + info["rollback_ip"] = info["rollback_host"]["ip"] + info["bk_rollback"] = info.pop("rollback_host") else: info["rollback_cluster_id"] = info.pop("target_cluster_id") @@ -122,12 +125,12 @@ def format(self): def post_callback(self): next_flow = self.ticket.next_flow() for info in next_flow.details["ticket_data"]["infos"]: - info["rollback_ip"] = info["rollback_host"]["ip"] - info["bk_rollback"] = info.pop("rollback_host") + info["rollback_ip"] = info["rollback_host"][0]["ip"] + info["bk_rollback"] = info.pop("rollback_host")[0] next_flow.save(update_fields=["details"]) -@builders.BuilderFactory.register(TicketType.MYSQL_ROLLBACK_CLUSTER) +@builders.BuilderFactory.register(TicketType.MYSQL_ROLLBACK_CLUSTER, is_apply=True) class MysqlFixPointRollbackFlowBuilder(BaseMySQLTicketFlowBuilder): serializer = MySQLFixPointRollbackDetailSerializer inner_flow_builder = MySQLFixPointRollbackFlowParamBuilder diff --git a/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_cluster.py b/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_cluster.py index e5802c2b78..7c2e0c546d 100644 --- a/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_cluster.py +++ b/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_cluster.py @@ -8,6 +8,7 @@ an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. """ +import itertools from django.utils.translation import gettext_lazy as _ from rest_framework import serializers @@ -82,14 +83,18 @@ def format_ticket_data(self): class MysqlMigrateClusterResourceParamBuilder(BaseOperateResourceParamBuilder): def format(self): - self.patch_info_affinity_location(roles=["backend_group"]) + self.patch_info_affinity_location() def post_callback(self): next_flow = self.ticket.next_flow() ticket_data = next_flow.details["ticket_data"] for info in ticket_data["infos"]: - backend = info.pop("backend_group")[0] - info["bk_new_master"], info["bk_new_slave"] = backend["master"], backend["slave"] + # 兼容资源池手动选择和自动匹配的协议 + if "backend_group" in info: + backend = info.pop("backend_group")[0] + info["bk_new_master"], info["bk_new_slave"] = backend["master"], backend["slave"] + else: + info["bk_new_master"], info["bk_new_slave"] = info.pop("new_master")[0], info.pop("new_slave")[0] info["new_master_ip"], info["new_slave_ip"] = info["bk_new_master"]["ip"], info["bk_new_slave"]["ip"] next_flow.save(update_fields=["details"]) @@ -103,17 +108,18 @@ class MysqlMigrateClusterFlowBuilder(MysqlMasterSlaveSwitchFlowBuilder): need_patch_recycle_host_details = True @staticmethod - def get_old_master_slave_host(info): - # 同机关联情况下，任取一台集群 - cluster = Cluster.objects.get(id=info["cluster_ids"][0]) - master = cluster.storageinstance_set.get(instance_inner_role=InstanceInnerRole.MASTER) - slave = cluster.storageinstance_set.get(instance_inner_role=InstanceInnerRole.SLAVE, is_stand_by=True) - # 补充下架的机器信息 - info["old_nodes"] = {"old_master": [master.machine.simple_desc], "old_slave": [slave.machine.simple_desc]} - return info + def get_old_master_slave_host(infos, cluster_map): + for info in infos: + # 同机关联情况下，任取一台集群 + insts = cluster_map[info["cluster_ids"][0]].storageinstance_set.all() + master = next(i for i in insts if i.instance_inner_role == InstanceInnerRole.MASTER) + slave = next(i for i in insts if i.instance_inner_role == InstanceInnerRole.SLAVE and i.is_stand_by) + # 补充下架的机器信息 + info["old_nodes"] = {"old_master": [master.machine.simple_desc], "old_slave": [slave.machine.simple_desc]} def patch_ticket_detail(self): + cluster_ids = list(itertools.chain(*[infos["cluster_ids"] for infos in self.ticket.details["infos"]])) + cluster_map = Cluster.objects.prefetch_related("storageinstance_set").in_bulk(cluster_ids, field_name="id") # mysql主从迁移会下架掉master和slave(stand by) - for info in self.ticket.details["infos"]: - self.get_old_master_slave_host(info) + self.get_old_master_slave_host(self.ticket.details["infos"], cluster_map) super().patch_ticket_detail() diff --git a/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_upgrade.py b/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_upgrade.py index 7bb7d506b9..43a3547581 100644 --- a/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_upgrade.py +++ b/dbm-ui/backend/ticket/builders/mysql/mysql_migrate_upgrade.py @@ -97,12 +97,37 @@ def format_ticket_data(self): class MysqlMigrateUpgradeResourceParamBuilder(BaseOperateResourceParamBuilder): + def auto_patch_info(self, info, info_index, nodes, cluster): + info["new_master"] = nodes[f"{info_index}_backend_group"][0]["master"] + info["new_slave"] = nodes[f"{info_index}_backend_group"][0]["slave"] + info["ro_slaves"] = [ + { + "old_ro_slave": { + "bk_cloud_id": slave.machine.bk_cloud_id, + "bk_host_id": slave.machine.bk_host_id, + "ip": slave.machine.ip, + }, + "new_ro_slave": nodes[f"{info_index}_{slave.machine.bk_host_id}"][0], + } + for slave in cluster.storageinstance_set.all() + if slave.instance_role == InstanceRole.BACKEND_SLAVE and not slave.is_stand_by + ] + + def manual_patch_info(self, info, info_index, cluster, nodes): + info["new_master"] = info["new_master"][0] + info["new_slave"] = info["new_slave"][0] + info["ro_slaves"] = [ + {"old_ro_slave": slave["old_slave"], "new_ro_slave": slave["new_slave"]} + for slave in info.pop("read_only_slaves", []) + ] + # 弹出read_only_new_slave，这个key仅作资源池申请 + info.pop("read_only_new_slave") + def post_callback(self): # 通过资源池获取到的节点 nodes = self.ticket_data.pop("nodes", []) cluster_ids = list(itertools.chain(*[infos["cluster_ids"] for infos in self.ticket.details["infos"]])) - id_cluster_map = Cluster.objects.prefetch_related( "storageinstance_set", "storageinstance_set__machine" ).in_bulk(cluster_ids, field_name="id") @@ -112,21 +137,10 @@ def post_callback(self): ticket_data = next_flow.details["ticket_data"] for info_index, info in enumerate(ticket_data["infos"]): + # 兼容资源池手动输入和自动匹配的协议 cluster = id_cluster_map[info["cluster_ids"][0]] - info["new_master"] = nodes[f"{info_index}_backend_group"][0]["master"] - info["new_slave"] = nodes[f"{info_index}_backend_group"][0]["slave"] - info["ro_slaves"] = [ - { - "old_ro_slave": { - "bk_cloud_id": slave.machine.bk_cloud_id, - "bk_host_id": slave.machine.bk_host_id, - "ip": slave.machine.ip, - }, - "new_ro_slave": nodes[f"{info_index}_{slave.machine.bk_host_id}"][0], - } - for slave in cluster.storageinstance_set.all() - if slave.instance_role == InstanceRole.BACKEND_SLAVE and not slave.is_stand_by - ] + # self.auto_patch_info(info, info_index, nodes, cluster) + self.manual_patch_info(info, info_index, cluster, nodes) ticket_data["infos"][info_index] = info next_flow.save(update_fields=["details"]) @@ -141,21 +155,12 @@ class MysqlMigrateUpgradeFlowBuilder(MysqlMasterSlaveSwitchFlowBuilder): resource_batch_apply_builder = MysqlMigrateUpgradeResourceParamBuilder need_patch_recycle_host_details = True - def patch_ticket_detail(self): - """mysql_master -> backend_group""" - # 主从构成 backend group - # 只读从库（非 standby） 各自单独成组 - super().patch_ticket_detail() - + def patch_auto_match_resource_spec(self, id_cluster_map): + # 自动匹配补充规格信息 resource_spec = {} - cluster_ids = list(itertools.chain(*[infos["cluster_ids"] for infos in self.ticket.details["infos"]])) - id_cluster_map = Cluster.objects.prefetch_related( - "storageinstance_set", "storageinstance_set__machine" - ).in_bulk(cluster_ids, field_name="id") - for info in self.ticket.details["infos"]: - cluster = id_cluster_map[info["cluster_ids"][0]] # 主从规格 + cluster = id_cluster_map[info["cluster_ids"][0]] ins = cluster.storageinstance_set.first() resource_spec["backend_group"] = { "spec_id": ins.machine.spec_id, @@ -172,8 +177,33 @@ def patch_ticket_detail(self): "location_spec": {"city": cluster.region, "sub_zone_ids": [ins.machine.bk_sub_zone_id]}, "affinity": AffinityEnum.NONE.value, } + info["old_nodes"]["old_slave"].append(ins.machine.simple_desc) + # 覆写resource_spec info["resource_spec"] = resource_spec - # 补充下架机器的信息 - MysqlMigrateClusterFlowBuilder.get_old_master_slave_host(info) - self.ticket.save(update_fields=["details"]) + def patch_manual_match_resource_spec(self, id_cluster_map): + # 手动匹配补充规格信息 + for info in self.ticket.details["infos"]: + read_only_new_slave = [slave["new_slave"] for slave in info["read_only_slaves"]] + read_only_old_slave = [slave["old_slave"] for slave in info["read_only_slaves"]] + info["old_nodes"]["old_slave"].extend(read_only_old_slave) + info["resource_spec"]["read_only_new_slave"] = {"spec_id": 0, "hosts": read_only_new_slave} + + def patch_ticket_detail(self): + """mysql_master -> backend_group""" + # 主从构成 backend group + # 只读从库（非 standby） 各自单独成组 + + cluster_ids = list(itertools.chain(*[infos["cluster_ids"] for infos in self.ticket.details["infos"]])) + id_cluster_map = Cluster.objects.prefetch_related( + "storageinstance_set", "storageinstance_set__machine" + ).in_bulk(cluster_ids, field_name="id") + + # 补充下架机器的信息 + MysqlMigrateClusterFlowBuilder.get_old_master_slave_host(self.ticket.details["infos"], id_cluster_map) + # 补充自动匹配的资源池信息 + # self.patch_auto_match_resource_spec(id_cluster_map) + # 兼容方案，先走资源池手动匹配协议 + self.patch_manual_match_resource_spec(id_cluster_map) + # 补充通用单据信息 + super().patch_ticket_detail() diff --git a/dbm-ui/backend/ticket/builders/redis/base.py b/dbm-ui/backend/ticket/builders/redis/base.py index 69fd4a341d..30621bd290 100644 --- a/dbm-ui/backend/ticket/builders/redis/base.py +++ b/dbm-ui/backend/ticket/builders/redis/base.py @@ -22,7 +22,7 @@ from backend.ticket import builders from backend.ticket.builders import TicketFlowBuilder from backend.ticket.builders.common.base import RedisTicketFlowBuilderPatchMixin, SkipToRepresentationMixin -from backend.ticket.constants import CheckRepairFrequencyType, DataCheckRepairSettingType +from backend.ticket.constants import CheckRepairFrequencyType, DataCheckRepairSettingType, FlowType KEY_FILE_PREFIX = "/redis/keyfiles/{biz}" @@ -120,6 +120,7 @@ class DataCheckRepairSettingSerializer(serializers.Serializer): class RedisUpdateApplyResourceParamBuilder(builders.ResourceApplyParamBuilder): def post_callback(self): next_flow = self.ticket.next_flow() + drop_proxy_hosts = [] for info in next_flow.details["ticket_data"]["infos"]: group_num = info["resource_spec"]["backend_group"]["count"] shard_num = info["cluster_shard_num"] @@ -139,8 +140,22 @@ def post_callback(self): # 分片数 shard_num=shard_num, ) + # 新proxy也会下架，这里需要加入到old_nodes里面 + drop_proxy_hosts.extend(info["proxy"]) + next_flow.save(update_fields=["details"]) + # 将下架的新proxy更新到清理流程中 + recycle_flow = self.ticket.flows.get(flow_type=FlowType.HOST_RECYCLE) + recycle_flow.details["ticket_data"]["clear_hosts"].extend(drop_proxy_hosts) + recycle_flow.save(update_fields=["details"]) + + # 如果有导入资源池流程，则将新proxy加入 + resource_flow = self.ticket.flows.filter(flow_type=FlowType.HOST_IMPORT_RESOURCE).first() + if resource_flow: + resource_flow.details["ticket_data"]["hosts"].extend(drop_proxy_hosts) + resource_flow.save(update_fields=["details"]) + class ClusterValidateMixin(object): """全局校验cluster状态""" diff --git a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_cut_off.py b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_cut_off.py index a334d7bbcc..a9dcd7ed7e 100644 --- a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_cut_off.py +++ b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_cut_off.py @@ -56,7 +56,6 @@ class RedisClusterCutOffParamBuilder(builders.FlowParamBuilder): controller = RedisController.redis_cluster_cutoff_scene def format_ticket_data(self): - self.ticket_data.pop("old_nodes") super().format_ticket_data() diff --git a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_proxy_scale_down.py b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_proxy_scale_down.py index b3c9ca6ef7..be8c2ce0c0 100644 --- a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_proxy_scale_down.py +++ b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_proxy_scale_down.py @@ -37,7 +37,6 @@ class OldProxySerializer(serializers.Serializer): cluster_id = serializers.IntegerField(help_text=_("集群ID")) target_proxy_count = serializers.IntegerField(help_text=_("目标proxy数量"), min_value=2, required=False) - proxy_reduce_count = serializers.IntegerField(help_text=_("缩容proxy数量"), required=False) old_nodes = OldProxySerializer(help_text=_("缩容指定proxy"), required=False) online_switch_type = serializers.ChoiceField( help_text=_("切换类型"), choices=SwitchConfirmType.get_choices(), default=SwitchConfirmType.NO_CONFIRM @@ -47,19 +46,18 @@ class OldProxySerializer(serializers.Serializer): ip_recycle = HostRecycleSerializer(help_text=_("主机回收信息"), default=HostRecycleSerializer.DEFAULT) def validate(self, attrs): - clusters = Cluster.objects.filter(id__in=fetch_cluster_ids(attrs)).prefetch_related("proxyinstance_set") - cluster_id__cluster_map = {cluster.id: cluster for cluster in clusters} - + cluster_ids = fetch_cluster_ids(attrs) + cluster_map = Cluster.objects.prefetch_related("proxyinstance_set").in_bulk(cluster_ids, field_name="id") # 验证缩容后数量至少为2 for info in attrs["infos"]: - cluster = cluster_id__cluster_map[info["cluster_id"]] - if info.get("proxy_reduced_hosts"): - info["target_proxy_count"] = cluster.proxyinstance_set.count() - len(info["proxy_reduced_hosts"]) + cluster = cluster_map[info["cluster_id"]] + if info.get("old_nodes"): + proxy_reduced_hosts = info["old_nodes"]["proxy_reduced_hosts"] + info["target_proxy_count"] = cluster.proxyinstance_set.count() - len(proxy_reduced_hosts) if info["target_proxy_count"] < 2: raise serializers.ValidationError(_("请保证集群{}缩容后proxy数量不小于2").format(cluster.immute_domain)) # 提前存入proxy信息用于后续patch - attrs.update(proxy_insts=cluster.proxyinstance_set.all(), bk_cloud_id=cluster.bk_cloud_id) - + attrs.update(bk_cloud_id=cluster.bk_cloud_id) return attrs @@ -67,7 +65,8 @@ class ProxyScaleDownParamBuilder(builders.FlowParamBuilder): controller = RedisController.redis_proxy_scale def format_ticket_data(self): - super().format_ticket_data() + for info in self.ticket_data["infos"]: + info["proxy_reduced_hosts"] = info["old_nodes"]["proxy_reduced_hosts"] @builders.BuilderFactory.register(TicketType.REDIS_PROXY_SCALE_DOWN, is_recycle=True) @@ -78,20 +77,23 @@ class ProxyScaleDownFlowBuilder(BaseRedisTicketFlowBuilder): need_patch_recycle_host_details = True def patch_old_proxy_nodes(self): + cluster_ids = fetch_cluster_ids(self.ticket.details) + cluster_map = Cluster.objects.prefetch_related("proxyinstance_set").in_bulk(cluster_ids, field_name="id") for info in self.ticket.details["infos"]: - proxy_insts = info.pop("proxy_insts") - if info.get("old_nodes"): continue - # 获取proxy ip和ip与host id的映射 - proxy_ip__host = {proxy.machine.ip: proxy.machine for proxy in proxy_insts} + proxy_insts = cluster_map[info["cluster_id"]].proxyinstance_set.all() + proxy_ip__host = {proxy.machine.ip: proxy.machine.bk_host_id for proxy in proxy_insts} proxy_ips = list(proxy_insts.values_list("machine__ip", flat=True)) # 获取实际下架的ip target_proxy_count = info["target_proxy_count"] down_ips = RedisProxyScaleFlow.calc_scale_down_ips(self.ticket.bk_biz_id, proxy_ips, target_proxy_count) # 补充old proxy nodes信息 - info["old_nodes"] = {"proxy": [{"bk_host_id": proxy_ip__host[ip], "ip": ip} for ip in down_ips]} + info["old_nodes"] = { + "proxy_reduced_hosts": [{"bk_host_id": proxy_ip__host[ip], "ip": ip} for ip in down_ips] + } + info.pop("target_proxy_count") def patch_ticket_detail(self): self.patch_old_proxy_nodes() diff --git a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_shard_update.py b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_shard_update.py index ebc63fc61b..af5b2a2676 100644 --- a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_shard_update.py +++ b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_shard_update.py @@ -108,3 +108,6 @@ class RedisShardUpdateFlowBuilder(BaseRedisTicketFlowBuilder): inner_flow_name = _("Redis 集群分片变更") resource_batch_apply_builder = RedisShardUpdateResourceParamBuilder need_patch_recycle_cluster_details = True + + def patch_recycle_cluster_details(self): + super().patch_recycle_cluster_details(role="backend") diff --git a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_type_update.py b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_type_update.py index e3df1fef7a..e6e1bd05e4 100644 --- a/dbm-ui/backend/ticket/builders/redis/redis_toolbox_type_update.py +++ b/dbm-ui/backend/ticket/builders/redis/redis_toolbox_type_update.py @@ -110,4 +110,7 @@ class RedisTypeUpdateFlowBuilder(BaseRedisTicketFlowBuilder): inner_flow_builder = RedisTypeUpdateParamBuilder inner_flow_name = _("Redis 集群类型变更") resource_batch_apply_builder = RedisTypeUpdateResourceParamBuilder - need_patch_recycle_cluster_details = True + need_patch_cluster_details = True + + def patch_recycle_cluster_details(self): + super().patch_recycle_cluster_details(role="backend") diff --git a/dbm-ui/backend/ticket/models/ticket.py b/dbm-ui/backend/ticket/models/ticket.py index 8225e2c392..043a0fe900 100644 --- a/dbm-ui/backend/ticket/models/ticket.py +++ b/dbm-ui/backend/ticket/models/ticket.py @@ -229,6 +229,8 @@ def create_recycle_ticket(cls, ticket_id: int, ip_dest: IpDest): "ip_recycle": {"ip_dest": ip_dest, "for_biz": ticket.bk_biz_id}, "group": ticket.group, } + if not details["recycle_hosts"]: + return recycle_ticket = cls.create_ticket( ticket_type=TicketType.RECYCLE_HOST, creator=ticket.creator,