incredible slow auto index

[lambre...@gmail.com]

hi,

i have a very small test data set (~2000 nodes and ~12000 relationships)

my node auto index is handling among others these properties:

id,type


my queries normally behave quite well when i look up like this:

start m = node:node_auto_index('type:player') RETURN DISTINCT count(m);

+----------+
| count(m) |
+----------+
| 1171 |
+----------+
1 row
17 ms



however today i found a pretty bad performance on this one:

start m = node:node_auto_index('type:referee') RETURN DISTINCT count(m);

+----------+
| count(m) |
+----------+
| 30 |
+----------+
1 row
22643 ms



i presume this is a lucene matter, but i just don't understand why it
would happen.



furthermore if i alter this slow query to include a wildcard like this:

start m = node:node_auto_index('type:referee AND id:*') RETURN DISTINCT
count(m);


+----------+
| count(m) |
+----------+
| 30 |
+----------+
1 row
10 ms


i get an instant response ?! so currently that's my fix.

all the above queries behave the same way repeatedly.

the server is running with neo4j-community-1.8.M06 with default config
in a linux environment


thanks
./allan

[Michael Hunger]

Can you just
RETURN COUNT(*)

How many entries are in that index?

Any chance of profiling it? Or sharing the graph data as a zip?

Sent from mobile device

> --
>
>

[Michael Hunger]

I checked your store copy and have the first query return after 6-10ms

and the second one (that's slow in your case) after 1-2ms.
Not sure what happend there for you.

(running neo4j 1.8.M07)

Michael

> --
>
>