# Operation Tool
## Overview
OpenMLDB provides an operation tool for user to use & maintain OpenMLDB conveniently. It contains the following ability:
- Recover data automatically: `recoverdata`
- Balance partitions when scale-out: `scaleout`
- Migrate partitions from specified endpoints to others when scale-in: `scalein`
- Pre-upgrade and post-upgrade for tablet upgrade: `pre-upgrade` and `post-upgrade`
- Check op status: `showopstatus`
- Check table status: `showtablestatus`
## Usage
### Common parameters
- --openmldb_bin_path: specify openmldb binary path
- --zk_cluster: zookeeper address
- --zk_root_path: zookeeper path
- --cmd: the operation to execute
### Operations
| Operation | Function | Extra Parameters |
|-------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| `recoverdata` | recover all tables | - |
| `scaleout` | rebalance partitions among tablets | - |
| `scalein` | migrate partitions from specified endpoints to others. If the number of remaining tablets is smaller than the max replicanum, it will fail | --endpoints: the endpoints that are migrated from, multiple endpoints are delimited by `,` |
| `pre-upgrade` | tablet pre-upgrade, which will migrate all the leader partitions to other tablets to reduce the influence on the online service. If there is only one replica for a partition, we will create an extra one which will be deleted automatically during `post-upgrade`. All the changes will be recorded in `statfile` | --endpoints: the endpoints that are upgraded, only one endpoint is allowed
--statfile:the temporary file to record the leader change information, which defaults to `.stat` |
| `post-upgrade` | tablet post-upgrade, which will revert all the changes recorded in `statfile` | --endpoints: the endpoints that are upgraded, only one endpoint is allowed
--statfile:the temporary file to record the leader change information, which defaults to `.stat` |
| `showopstatus` | check the op status | --filter: only show the records with status `filter`. Valid statuses are `kInited`, `kDoing`, `kDone`, `kFailed` and `kCanceled`.Its default is `None`, meaning showing all the records |
| `showtablestatus` | check the table status | --filter: the required database pattern. Its matching rule is the same as `LIKE` op. Its default is `'%'`,meaning showing the tables under all the databases |
**Example**
```
python tools/openmldb_ops.py --openmldb_bin_path=./bin/openmldb --zk_cluster=0.0.0.0:2181 --zk_root_path=/openmldb --cmd=scaleout
python tools/openmldb_ops.py --openmldb_bin_path=./bin/openmldb --zk_cluster=0.0.0.0:2181 --zk_root_path=/openmldb --cmd=recoverdata
```
You can focus on whether there are ERROR-level logs in the running results. If there are, please keep the complete log records for technical personnel to investigate the issue.
### System Requirements
- python >= 2.7
- Note: In theory, `openmldb_ops` does not require version matching. A higher version of `openmldb_ops` can operate on a lower version of the `openmldb`
cluster.
- `showopstatus` and `showtablestatus` require `prettytable` dependency