Different shortest results in browser and bloom(2nd case)

Neo4j Graph Platform Neo4j Bloom
1

< Neo4j Browser >

match (m:Machine{name:'DC1-RCK-1-2-M-3'}), (m2:Machine{name:'DC1-RCK-1-4-M-45'})
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node))
return path
Started streaming 1 records after 6 ms and completed after 6 ms.

graph
graph968×137 18.6 KB

< Bloom(2.21) Scene Action >

match (m) where elementId(m) in $nodes
match (m2) where elementId(m2) in $nodes
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node))
return path
This took about 33 seconds!

bloom
bloom1203×385 59 KB

The two results should be the same, but they are not.
Bloom takes too long - there must be something wrong.

2

They shouldn't be the same ...

Query 1:

match (m:Machine{name:'DC1-RCK-1-2-M-3'}), (m2:Machine{name:'DC1-RCK-1-4-M-45'})
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node))
return path

Query 2 (modified):

:params { nodes: [ 'DC1-RCK-1-2-M-3','DC1-RCK-1-4-M-45' ] }

match (m) where m.name in $nodes
match (m2) where m2.name in $nodes
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node))
return path

This bit of your query: (m)-[:!TYPE&!HOLDS]-+(m2) combines the traversal m = M-3 -> M-45 and M45 -> 3 because your 2nd query is on the list in $nodes, while the first is on the node M3 only

Re:times - both queries were pants for me ... Q1 took 25 seconds and Q2 double at 50 secs - but i just ran it cold.

1 Like
3

Thank you for the kind explanations.
I modified the query as below, and it worked quickly and correctly:
match (m) where elementId(m) in $nodes
with collect(m) as nodes
CALL apoc.path.expandConfig(nodes[0], {
labelFilter: "-DataCenter|-Version|-Type",
endNodes: [nodes[1]],
minLevel: 2,
uniqueness: "RELATIONSHIP_PATH"
}) YIELD path
RETURN path LIMIT 1

If the query takes long time for you to run, adding an index on the name property will make it fast.

4

you don't need apoc ...

:params { nodes: [ 'DC1-RCK-1-2-M-3','DC1-RCK-1-4-M-45' ] }

WITH $nodes[0] AS node1, $nodes[1] AS node2
WITH node1, node2
match (m) where m.name = node1
match (m2) where m2.name = node2
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node))
return path
5

It doesn't work. $nodes in Bloom Scene Actions is the list of elementId of nodes selected on the scene.

6

When I was developing the query, I initially tried like
match (m) where elementId(m) = $nodes[0]
But it didn't work(I am not sure why).
However "match (m) where elementId(m) in $nodes" works, so I used that before.

1 Like
7

I think the query below should work, but it doesn't:

match (m) where elementId(m) in $nodes 
with collect(m) as nodes
with nodes[0] as m, nodes[1] as m2
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node) or 'Type' in labels(node))
return path
8

what values do you have in $nodes?

:params { nodes: [ 'DC1-RCK-1-2-M-3','DC1-RCK-1-4-M-45' ] }

match (N) WHERE N.name in $nodes WITH collect(elementID(N)) as theIDs

WITH theIDs
match (m) where elementId(m) in theIDs
with collect(m) as nodes
with nodes[0] as m, nodes[1] as m2
match path = shortest 1 (m)-[:!TYPE&!HOLDS]-+(m2)
where not any(node in nodes(path) where 'DataCenter' in labels(node) or 'Version' in labels(node) or 'Type' inlabels(node))

return path
9

$nodes and $relationships are the default variables defined in Scene Actions of Bloom.

10

Since I don't know what you're using to load the variables - I am just showing you that the query works if you pre-load them.

11

What I mean is that $nodes is defined by Bloom as elemetId, not by me.

1 Like