feat: 定点回档物理备份changeMasterTEST3 TencentBlueKing#8338

zfrendo · Dec 5, 2024 · 2cac206 · 2cac206
1 parent b648122
commit 2cac206
Showing 1 changed file with 33 additions and 26 deletions.
diff --git a/dbm-ui/backend/flow/engine/bamboo/scene/mysql/mysql_rollback_data_flow.py b/dbm-ui/backend/flow/engine/bamboo/scene/mysql/mysql_rollback_data_flow.py
@@ -186,7 +186,8 @@ def rollback_data_flow(self):
                 backupinfo = get_local_backup(inst_list, cluster_class, mycluster["rollback_time"])
                 if backupinfo is None:
                     logger.error("cluster {} backup info not exists".format(cluster_class.id))
-                    raise TendbGetBackupInfoFailedException(message=_("获取集群 {} 的备份信息失败".format(cluster_class.id)))
+                    raise TendbGetBackupInfoFailedException(
+                        message=_("获取集群 {} 的备份信息失败".format(cluster_class.id)))
                 mycluster["backupinfo"] = copy.deepcopy(backupinfo)
 
                 sub_pipeline.add_sub_pipeline(
@@ -278,20 +279,48 @@ def rollback_to_cluster_flow(self):
                 db_module_id=self.data["db_module_id"],
                 cluster_type=self.data["cluster_type"],
             )
+            # 为了回档的备份信息各个节点统一，获取备份的信息提前
+            backupinfo = copy.deepcopy(self.data["backupinfo"])
+            if self.data["rollback_type"] == RollbackType.LOCAL_AND_TIME:
+                # 可从主库获取,但需要排除掉被回档的机器
+                inst_list = ["{}{}{}".format(master.machine.ip, IP_PORT_DIVIDER, master.port)]
+                stand_by_slaves = cluster_class.storageinstance_set.filter(
+                    instance_inner_role=InstanceInnerRole.SLAVE.value,
+                    is_stand_by=True,
+                    status=InstanceStatus.RUNNING.value,
+                )
+                if len(stand_by_slaves) > 0:
+                    inst_list.append(
+                        "{}{}{}".format(stand_by_slaves[0].machine.ip, IP_PORT_DIVIDER, stand_by_slaves[0].port)
+                    )
+                backupinfo = get_local_backup(inst_list, cluster_class, self.data["rollback_time"])
+                if backupinfo is None:
+                    logger.error("cluster {} backup info not exists".format(cluster_class.id))
+                    raise TendbGetBackupInfoFailedException(
+                        message=_("获取集群 {} 的备份信息失败".format(cluster_class.id)))
+
+            elif self.data["rollback_type"] == RollbackType.REMOTE_AND_TIME.value:
+                rollback_handler = FixPointRollbackHandler(cluster_class.id)
+                backupinfo = rollback_handler.query_latest_backup_log(str2datetime(self.data["rollback_time"]))
+                if backupinfo is None:
+                    logger.error("cluster {} backup info not exists".format(["cluster_id"]))
+                    raise TendbGetBackupInfoFailedException(
+                        message=_("获取集群 {} 的备份信息失败".format(cluster_class.id))
+                    )
 
             sub_pipeline = SubBuilder(root_id=self.root_id, data=copy.deepcopy(self.data))
             rollback_class = Cluster.objects.get(id=self.data["rollback_cluster_id"])
             storages = rollback_class.storageinstance_set.all()
             rollback_pipeline_list = []
             for rollback_storage in storages:
                 if not check_storage_database(
-                    rollback_class.bk_cloud_id, rollback_storage.machine.ip, rollback_storage.port
+                        rollback_class.bk_cloud_id, rollback_storage.machine.ip, rollback_storage.port
                 ):
                     logger.error("cluster {} check database fail".format(rollback_class.id))
                     raise NormalSpiderFlowException(
-                        message=_("回档集群 {} 空闲检查不通过，请确认回档集群是否存在非系统数据库".format(rollback_class.id))
+                        message=_(
+                            "回档集群 {} 空闲检查不通过，请确认回档集群是否存在非系统数据库".format(rollback_class.id))
                     )
-                backupinfo = copy.deepcopy(self.data["backupinfo"])
                 mycluster = {
                     "name": cluster_class.name,
                     "cluster_id": cluster_class.id,
@@ -369,20 +398,6 @@ def rollback_to_cluster_flow(self):
                     )
                 # 本地备份+时间
                 if self.data["rollback_type"] == RollbackType.LOCAL_AND_TIME:
-                    inst_list = ["{}{}{}".format(master.machine.ip, IP_PORT_DIVIDER, master.port)]
-                    stand_by_slaves = cluster_class.storageinstance_set.filter(
-                        instance_inner_role=InstanceInnerRole.SLAVE.value,
-                        is_stand_by=True,
-                        status=InstanceStatus.RUNNING.value,
-                    ).exclude(machine__ip__in=[rollback_storage.machine.ip])
-                    if len(stand_by_slaves) > 0:
-                        inst_list.append(
-                            "{}{}{}".format(stand_by_slaves[0].machine.ip, IP_PORT_DIVIDER, stand_by_slaves[0].port)
-                        )
-                    backupinfo = get_local_backup(inst_list, cluster_class, mycluster["rollback_time"])
-                    if backupinfo is None:
-                        logger.error("cluster {} backup info not exists".format(cluster_class.id))
-                        raise TendbGetBackupInfoFailedException(message=_("获取集群 {} 的备份信息失败".format(cluster_class.id)))
                     mycluster["backupinfo"] = copy.deepcopy(backupinfo)
                     rollback_pipeline.add_sub_pipeline(
                         sub_flow=rollback_local_and_time(
@@ -392,16 +407,8 @@ def rollback_to_cluster_flow(self):
                             cluster_model=cluster_class,
                         )
                     )
-
                 # 远程备份+时间
                 elif self.data["rollback_type"] == RollbackType.REMOTE_AND_TIME.value:
-                    rollback_handler = FixPointRollbackHandler(cluster_class.id)
-                    backupinfo = rollback_handler.query_latest_backup_log(str2datetime(mycluster["rollback_time"]))
-                    if backupinfo is None:
-                        logger.error("cluster {} backup info not exists".format(mycluster["cluster_id"]))
-                        raise TendbGetBackupInfoFailedException(
-                            message=_("获取集群 {} 的备份信息失败".format(mycluster["cluster_id"]))
-                        )
                     mycluster["backupinfo"] = copy.deepcopy(backupinfo)
                     rollback_pipeline.add_sub_pipeline(
                         sub_flow=rollback_remote_and_time(