Skip to content

Dump Tool

Dump Tool is a single-machine off-line data dumping tool that can be used to dump or count data with specified conditions.

How to Get

The source code of the dump tool is under nebula/src/tools/db_dump directory. You can use command make db_dump to compile it. Before using this tool, you can use the SHOW HOSTS statement in the Nebula Graph console to check the distribution of the partitions. Also, you can use the vertex_id % partition_num - 1 formula to calculate in which partition the vertex's corresponding key is located.

NOTE: The Dump Tool is located in the rpm package and its directory is nebula/bin/. Because this tool dumps data by directly opening the RocksDB, you need to use it offline. Stop the storaged process and make sure that the metad service is started. Please refer to the following section for detailed usage.

How to Use

The db_dump command displays information about how to use the dump tool. Parameter space is required. Parameters db_path and meta_server both have default values and you can configure them based on your actual situation. You can combine parameters vids, parts, tags and edges arbitrarily to dump the data you want.

  ./db_dump --space=<space name>

       --space=<space name>
        # A space name must be given.

       --db_path=<path to rocksdb>
         # Path to the RocksDB data directory. If nebula was installed in `/usr/local/nebula`,
         # the db_path would be /usr/local/nebula/data/storage/nebula/
         # Default: ./

         # A list of meta severs' ip:port separated by comma.
         # Default:

       --mode= scan | stat
         # scan: print to screen when records meet the condition, and also print statistics to screen in final.
         # stat: print statistics to screen.
         # Default: scan

       --vids=<list of vid>
         # A list of vids separated by comma. This parameter means vertex_id/edge_src_id
         # Would scan the whole space's records if it is not given.

       --parts=<list of partition id>
         # A list of partition ids separated by comma.
         # Would output all partitions if it is not given.

       --tags=<list of tag name>
         # A list of tag name separated by comma.

       --edges=<list of edge name>
         # A list of edge name separated by comma.

         # A positive number that limits the output.
         # Would output all if set to 0 or negative.
         # Default: 1000

Following is an example:

// Specify a space to dump data
./db_dump --space=space_name

// Specify space, db_path, meta_server
./db_dump --space=space_name --db_path=/usr/local/nebula/data/storage/nebula/ --meta_server=

// Set mode to stat, only stats are returned, no data is printed
./db_dump --space=space_name --mode=stat --db_path=/usr/local/nebula/data/storage/nebula/ --meta_server=

// Specify vid to dump the vertex and the edges sourcing from it
./db_dump --space=space_name --mode=stat --db_path=/usr/local/nebula/data/storage/nebula/ --meta_server= --vids=123,456

// Specify tag and dump vertices with the tag 
./db_dump --space=space_name --mode=stat --db_path=/usr/local/nebula/data/storage/nebula/ --meta_server= --tags=tag1,tag2

The returned data format:

// vertices, key: part_id, vertex_id, tag_name, value: <prop_list>
[vertex] key: 1, 0, poi value:mid:0,8191765721868409651,8025713627522363385,1993089399535188613,3926276052777355165,5123607763506443893,2990089379644866415,poi_name_0,上海,华东,30.2824,120.016,poi_stat_0,poi_fc_0,poi_sc_0,0,poi_star_0,

// edges, key: part_id, src_id, edge_name, ranking, dst_id, value: <prop_list>
[edge] key: 1, 0, consume_poi_reverse, 0, 656384 value:mid:656384,mid:0,7.19312,mid:656384,3897457441682646732,mun:656384,4038264117233984707,dun:656384,empe:656384,mobile:656384,gender:656384,age:656384,rs:656384,fpd:656384,0.75313,1.34433,fpd:656384,0.03567,7.56212,

// stats
COUNT: 10           # total number of data dumped
VERTEX COUNT: 1     # number of vertices dumped
EDGE COUNT: 9       # number of edges dumped
TAG STATISTICS:     # number of tags dumped
        poi : 1
EDGE STATISTICS:    # number of edge types dumped
        consume_poi_reverse : 9

Last update: April 8, 2021