ArangoDB v3.10 is under development and not released yet.
This documentation is not final and potentially incomplete.
To restore data from a dump previously created with arangodump, ArangoDB provides the arangorestore tool.
arangorestore can be invoked from the command-line as follows:
arangorestore --input-directory "dump"
This will connect to an ArangoDB server (tcp://127.0.0.1:8529 by default), then restore the collection structure and the documents from the files found in the input directory dump. Note that the input directory must have been created by running arangodump before.
arangorestore will by default connect to the _system database using the default endpoint. To override the endpoint, or specify a different user, use one of the following startup options:
--server.endpoint <string>: endpoint to connect to
--server.username <string>: username
--server.password <string>: password to use (omit this and you’ll be prompted for the password)
--server.authentication <bool>: whether or not to use authentication
If you want to connect to a different database or dump all databases you can additionally use the following startup options:
--server.database <string>: name of the database to connect to. Defaults to the
--all-databases true: restore multiple databases from a dump which used the same option. Introduced in v3.5.0.
Note that the specified user must have access to the database(s).
Since version 2.6 arangorestore provides the option --create-database. Setting this option to true will create the target database if it does not exist. When creating the target database, the username and passwords passed to arangorestore (in options --server.username and --server.password) will be used to create an initial user for the new database.
--force-same-database allows restricting arangorestore operations to a
database with the same name as in the source dump’s
dump.json file. It can thus be used
to prevent restoring data into a “wrong” database by accident.
For example, if a dump was taken from database A, and the restore is attempted into
database B, then with the
--force-same-database option set to
will abort instantly.
--force-same-database option is set to
false by default to ensure backwards-compatibility.
Here’s an example of reloading data to a non-standard endpoint, using a dedicated database name:
arangorestore \ --server.endpoint tcp://192.168.173.13:8531 \ --server.username backup \ --server.database mydb \ --input-directory "dump" \
Also, more than one endpoint can be provided, such as:
arangorestore \ --server.endpoint tcp://192.168.173.13:8531 \ --server.endpoint tcp://192.168.173.13:8532 \ --server.username backup \ --server.database mydb \ --input-directory "dump"
To create the target database when restoring, use a command like this:
arangorestore --server.username backup --server.database newdb --create-database true --input-directory "dump"
In contrast to the above calls, when working with multiple databases using
--server.database mydb must not be specified:
arangorestore --server.username backup --all-databases true --create-database true --input-directory "dump-multiple"
arangorestore will print out its progress while running, and will end with a line showing some aggregate statistics:
Processed 2 collection(s), read 2256 byte(s) from datafiles, sent 2 batch(es)
By default, arangorestore will re-create all non-system collections found in the input directory and load data into them. If the target database already contains collections which are also present in the input directory, the existing collections in the database will be dropped and re-created with the data found in the input directory.
The following parameters are available to adjust this behavior:
--create-collection <bool>: set to true to create collections in the target database. If the target database already contains a collection with the same name, it will be dropped first and then re-created with the properties found in the input directory. Set to false to keep existing collections in the target database. If set to false and arangorestore encounters a collection that is present in the input directory but not in the target database, it will abort. The default value is true.
--import-data <bool>: set to true to load document data into the collections in the target database. Set to false to not load any document data. The default value is true.
--include-system-collections <bool>: whether or not to include system collections when re-creating collections or reloading data. The default value is false.
For example, to (re-)create all non-system collections and load document data into them, use:
arangorestore --create-collection true --import-data true --input-directory "dump"
This will drop potentially existing collections in the target database that are also present in the input directory.
To include system collections too, use
arangorestore --create-collection true --import-data true --include-system-collections true --input-directory "dump"
To (re-)create all non-system collections without loading document data, use:
arangorestore --create-collection true --import-data false --input-directory "dump"
This will also drop existing collections in the target database that are also present in the input directory.
To just load document data into all non-system collections, use:
arangorestore --create-collection false --import-data true --input-directory "dump"
To restrict reloading to just specific collections, there is is the
It can be specified multiple times if required:
arangorestore --collection myusers --collection myvalues --input-directory "dump"
Collections will be processed in alphabetical order by arangorestore, with all document collections being processed before all edge collections. This remains valid also when multiple threads are in use (from v3.4.0 on).
Note however that when restoring an edge collection no internal checks are made in order to validate that the documents that the edges connect exist. As a consequence, when restoring individual collections which are part of a graph you are not required to restore in a specific order.
When restoring only a subset of collections of your database, and graphs are in use, you will need to make sure you are restoring all the needed collections (the ones that are part of the graph) as otherwise you might have edges pointing to non existing documents.
To restrict reloading to specific views, there is the
Should you specify the
--collection parameter views will not be restored unless you explicitly
specify them via the
arangorestore --collection myusers --view myview --input-directory "dump"
In the case of an arangosearch view you must make sure that the linked collections are either also restored or already present on the server.
See arangodump for details.
Reloading Data into a different Collection
arangorestore will restore document and edges data with the exact same _key, _rev, _from and _to values as found in the input directory.
With some creativity you can also use arangodump and arangorestore to transfer data from one collection into another (either on the same server or not). For example, to copy data from a collection myvalues in database mydb into a collection mycopyvalues in database mycopy, you can start with the following command:
arangodump --collection myvalues --server.database mydb --output-directory "dump"
This will create two files,
myvalues.data.json, in the output
directory. To load data from the datafile into an existing collection mycopyvalues in database
mycopy, rename the files to
After that, run the following command:
arangorestore --collection mycopyvalues --server.database mycopy --input-directory "dump"
Enabling revision trees for older dumps
Introduced in: v3.8.7, v3.9.2
Collections in ArangoDB 3.8 and later can use an internal format that is based on revision trees for replication. Using this format has advantages over the previous format, because changes to the collection on the leader can quickly be detected when trying to get follower shards in sync.
Dumps taken from older versions of ArangoDB, i.e. ArangoDB 3.7 or before, do not contain any information about revision trees. The arangorestore behavior for these collections is as follows:
- In ArangoDB versions before 3.8.7 and 3.9.2, the collections are restored without revision trees.
- In ArangoDB versions 3.8.7, 3.9.2 or later, the
collections use revision trees by default, but you can opt out of this by
invoking arangorestore with the
--enable-revision-trees startup option is
true (which is the default value),
then arangorestore adds the necessary attributes for using revision trees
when restoring the collections. It’s only done for the attributes which are not
contained in the dump. If the option is set to
false, arangorestore does not
add the attributes when restoring collections.
Regardless of the setting of this option, arangorestore does not add the attributes, when they are already present in the dump. You may modify the attributes manually in the dump if you want to change their values.
Restoring in a Cluster
To restore data into a Cluster, simply point arangorestore to one of the Coordinators in your Cluster.
If arangorestore is asked to restore a collection, it will use the same number of shards, replication factor and shard keys as when the collection was dumped. The distribution of the shards to the servers will also be the same as at the time of the dump, provided that the number of DB-Servers in the cluster dumped from is identical to the number of DB-Servers in the to-be-restored-to cluster.
To modify the number of shards or the replication factor for all or just
some collections, arangorestore provides the options
--replication-factor (starting from v3.3.22 and v3.4.2). These options
can be specified multiple times as well, in order to override the settings
for dedicated collections, e.g.
arangorestore --number-of-shards 2 --number-of-shards mycollection=3 --number-of-shards test=4
The above will restore all collections except “mycollection” and “test” with 2 shards. “mycollection” will have 3 shards when restored, and “test” will have 4. It is possible to omit the default value and only use collection-specific overrides. In this case, the number of shards for any collections not overridden will be determined by looking into the “numberOfShards” values contained in the dump.
--replication-factor options works in the same way, e.g.
arangorestore --replication-factor 2 --replication-factor mycollection=1
will set the replication factor to 2 for all collections but “mycollection”, which will get a replication factor of just 1.
replication-factor, as well as the deprecated
not applicable to system collections. They are managed by the server.
If a collection was dumped from a single instance and is then restored into
a cluster, the sharding will be done by the
_key attribute by default. One can
manually edit the structural description for the shard keys in the dump files if
If you restore a collection that was dumped from a cluster into a single ArangoDB instance, the number of shards, replication factor and shard keys will silently be ignored.
Factors affecting speed of arangorestore in a Cluster
The following factors affect speed of arangorestore in a Cluster:
- Replication Factor: the higher the replication factor, the more time the restore will take. To speed up the restore you can restore using a replication factor of 1 and then increase it again after the restore. This will reduce the number of network hops needed during the restore.
- Restore Parallelization: if the collections are not restored in
parallel, the restore speed is highly affected. A parallel restore can
be done from v3.4.0 by using the
--threadsoption of arangorestore. Before v3.4.0 it is possible to achieve parallelization by restoring on multiple Coordinators at the same time. Depending on your specific case, parallelizing on multiple Coordinators can still be useful even when the
--threadsoption is in use (from v.3.4.0).
- Dump Format: Since ArangoDB 3.8 arangodump can produce two different
dump formats: an enveloped format, which was the default format up to
including ArangoDB 3.8, and a non-envelop format, which is the default
since ArangoDB 3.9.0.
The enveloped format is downwards-compatible with all previous versions
of ArangoDB, and should only be used for dumps that need to be restored
into versions older than 3.9. The non-envelope format is only understood
since ArangoDB 3.8.0 and not compatible with previous versions. However, it
is smaller and slightly faster to produce. In addition, the non-envelope
format allows arangorestore to parallelize the restore operations not
only across collections but also within collections. The latter is not
possible with the envelope dump format.
In order to use the non-envelope dump format, invoke arangodump with the
--envelope false. arangorestore can automatically parallelize the restore of such dumps even for individual collections.
See Fast Cluster Restore for further operative details on how to take the three factors described above into account when restoring with arangorestore.
Restoring collections with sharding prototypes
arangorestore will yield an error when trying to restore a collection whose shard distribution follows a collection which does not exist in the cluster and which was not dumped along:
arangorestore --collection clonedCollection --server.database mydb --input-directory "dump" ERROR got error from server: HTTP 500 (Internal Server Error): ArangoError 1486: must not have a distributeShardsLike attribute pointing to an unknown collection Processed 0 collection(s), read 0 byte(s) from datafiles, sent 0 batch(es)
The collection can be restored by overriding the error message as follows:
arangorestore --collection clonedCollection --server.database mydb --input-directory "dump" --ignore-distribute-shards-like-errors
Restore into an authentication-enabled ArangoDB
Of course you can restore data into a password-protected ArangoDB as well. However this requires certain user rights for the user used in the restore process. The rights are described in detail in the Managing Users chapter. For restore this short overview is sufficient:
- When importing into an existing database, the given user needs
Administrateaccess on this database.
- When creating a new database during restore, the given user needs
_system. The user will be promoted with
Administrateaccess on the newly created database.