Create new database using neo4j-admin or cypher-shell

Neo4j Graph Platform Cypher
1

Hi
I do not have access to the neo4j UI console due to ports mis-configurations in kubernetes, however I am able to import files and load files from a nodes and relationships files in a db called userdb. I can also see the folder in databases dir.

However to use this when i try to use cypher-shell for creating the database it does not seem to work.
With cypher-shell "Create dababase userdb" resulting in
Upsupported administration command: create database userdb

I have searched and neo4j-admin also does not have the create command.
I would appreciate any help.
Thanks in advance
Thanks

2

@RikkiGraf

what version of Neo4j?
Are you using Community Edition? which only supports 1 database.

3

Enterprise version
5.19.0

4

@RikkiGraf

at time of error anything logged in logs\debug.log?

5

Not seeing anything in the debug.log file. just ran the command again and same error!!

6

Just for completeness' sake - what do you get when you run:

call dbms.components() yield name, versions, edition unwind versions as version return name, version, edition;
7

Open your Neo4j browser and click on the database icon at top left. When clicked you will see a dropdown list of databases. In your case you will see 'neo4j' and 'system'. select 'system' . Run the query "SHOW DATABASES". This will show all the databases . Execute the Cypher query: CREATE DATABASE userdb. Once successful, rerun the SHOW DATABASES and now you should see you db name. Also check to see if the db is active. Now in the left db dropdown select your db.

8

Thanks , but as I mentioned I dont have access to the UI. so using the cypher-shell and neo4j-admin.

9

ok turns out it is "Neo4j Kernel" | 5.23.0" | "community" , the IT ended up doing a community installation.
So now my question is

  • how do i use the only database "neo4j" in community edition to load/import nodes.csv and edges.csv using the cypher-shell/neo4j-admin tool ?
    Thanks
10

Have you tried LOAD CSV - Cypher Manual

11

./bin/neo4j-admin database import full neo4j —nodes=/import/nodes.csv —relationships=/import/edges.csv —trim-strings=true

This is the command that I have to load into the neo4j pods, but when I tried to execute it in the neo4j (which is running in kubernetes pods) it asked me to shut down the server. When I did that it kicked me out of the pods, wondering now where should I run the neo4j-admin command

If I were to use LOAD CSV which I have used in the past but requires me to understand the schema and appropriately add them into my LOAD CSV where I want to make connections etc.

However in this case this is an external dataset and please suggest how I go about loading it.

Thanks

12

There is an offline maintenance mode provided for this purpose:

From one of my colleagues:

You can run a helm-upgrade to put the pod in maintenance mode:

helm upgrade neo4j --set neo4j.offlineMaintenanceModeEnabled=true

Once applied, the pod will restart but Neo4j won't be running. Instead, it will run a dummy process that outputs a message indicating it's in maintenance mode.
Then you can exec into the pod and perform your offline import.

13

This is what I have ended up using for loading the nodes:

LOAD CSV WITH HEADERS FROM "file:///nodes_load.csv" AS row
CALL apoc.create.node([row.LABEL],{node_id:row.node_id, node_index:row.node_index})
yield node
return count(*)

And it was quick,
However when loading the relationships which are about 5M:

CALL apoc.periodic.iterate(
'CALL apoc.load.csv("file:///edges_small.csv") yield map AS row return row',
'MATCH (n{node_index:row.START_ID}), (m{node_index:row.END_ID})
CALL apoc.create.relationship(n, row.TYPE, {relation:row.disrelation},m)
yield rel return rel',
{batchSize:50000, parallel:true})

It has been very slow , has been going on for several hours and only 400k relations are loaded.
Any pointers for speeding it up?

14

Do you have indexes on START_ID and END_ID? Not having those (indexes to enable you to find the relationship ends quickly) is almost always the cause of poor performance loading relationships.

15

I have 5 different types of nodes with node_index property and I have constraint set on all 5 of those nodes:node_index!!

16

Can you please return the output of cypher

show indexes;

Not that I'm suggesting apoc is the cause of your performance issue but why complicate matters with apoc when this could have been without the need for apoc and as described at

17

I had the constraints set on all the source and target properties, so assuming that would create the indexes as well.
I let the script run over the weekend and it took about 36 hours for 130k nodes and 4.5M relationships.
Thanks for all the help

18

It is documented that creating property uniqueness or key constraints will create indexes as well. Having said that, it still seems pretty slow - did you try using LOAD CSV directly, without involving apoc?