## Proposed changes
The current implement will persist all catalogs/databases of external catalogs, and only the master FE can handle hms events and make all slave nodes replay these events, this will bring some problems:
- The hms event processor ( `MetastoreEventsProcessor` ) can not consume the events in time. (Add journal log is a synchronized method, we can not speed up the consume rate by using concurrent processing, and each add-journal-log operation costs about tens of milliseconds) So the meta info of hive maybe out of date.
- Slave FE nodes maybe crashed if FE replays the journal logs of hms events failed. (In fact we have fixed some issues about this, but we can not make sure all the issues have been resolved)
- There are many journal logs which are produced by hms events, but in fact these logs are not used anymore after FE restart. It makes the start time of all FE nodes very long.
Now doris try to persis all databases/tables of external catalogs just to make sure that the dbId/tableId of databases/tables are the same through all FE nodes, it will be used by analysis jobs.
In this pr, we use a meta id manager called `ExternalMetaIdMgr` to manage these meta ids. On every loop when master fetches a batch of hms events, it handles the meta ids first and produce only one meta id mappings log, slave FE nodes will replay this log to sync the changes about these meta ids. `MetastoreEventsProcessor` will start on every FE nodes and try to consume these hms events as soon as possible.
## Further comments
I've submitted two prs ( #22869#21589 ) to speed up the consume rate of hms events before, it works fine when there are many `AlterTableEvent` / `DropTableEvent` on hive cluster. But the improvement is not that significant when most of hms events are partition-events. Unfortunately, we performed a cluster upgrade (upgrade spark 2.x to spark 3.x), maybe this is the reason that resulting in the majority of Hive Metastore events became partition-events. This is also the reason for the existence of this pull request.
Based on our observation, after merging this pull request, Doris is now capable of processing thousands of Hive Metastore events per second, compared to the previous capability of handling only a few dozen events.
```java
2023-12-07 05:17:03,518 INFO (replayer|105) [Env.replayJournal():2614] replayed journal id is 18287902, replay to journal id is 18287903
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEventFactory.mergeEvents():188] Event size on catalog [xxx] before merge is [1947], after merge is [1849]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955309 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[partitions=2022-05-27],partitionNameAfter:[partitions=2022-05-27]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955310 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[pday=20230318],partitionNameAfter:[pday=20230318]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955311 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[pday=20190826],partitionNameAfter:[pday=20190826]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955312 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[partitions=2021-09-16],partitionNameAfter:[partitions=2021-09-16]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955314 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[partitions=2020-04-26],partitionNameAfter:[partitions=2020-04-26]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955315 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[pday=20230702],partitionNameAfter:[pday=20230702]
2023-12-07 05:17:03,735 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357955317 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[pday=20211019],partitionNameAfter:[pday=20211019]
...
2023-12-07 05:17:03,989 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357957252 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[partitions=2021-08-27],partitionNameAfter:[partitions=2021-08-27]
2023-12-07 05:17:03,989 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEvent.infoLog():193] EventId: 357957253 EventType: ALTER_PARTITION catalogName:[xxx],dbName:[xxx],tableName:[xxx],partitionNameBefore:[partitions=2022-02-05],partitionNameAfter:[partitions=2022-02-05]
2023-12-07 05:17:04,661 INFO (replayer|105) [Env.replayJournal():2614] replayed journal id is 18287903, replay to journal id is 18287904
2023-12-07 05:17:05,028 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEventsProcessor.realRun():116] Events size are 587 on catalog [xxx]
2023-12-07 05:17:05,662 INFO (org.apache.doris.datasource.hive.event.MetastoreEventsProcessor|37) [MetastoreEventFactory.mergeEvents():188] Event size on catalog [xxx] before merge is [587], after merge is [587]
```
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements. See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership. The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License. You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on 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.
# fe-common
This module is used to store some common classes of other modules.
# spark-dpp
This module is Spark DPP program, used for Spark Load function.
Depends: fe-common
# fe-core
This module is the main process module of FE.
Depends: fe-common, spark-dpp
8a75da0fec