Neo4j 3.3.0 community edition Open NLP - neo4j

Installed Neo4j CE 3.3.0 on windows RAM 8GB. I referred to https://medium.com/#david.allen_3172/using-nlp-in-neo4j-ac40bc92196f for installation of the open NLP and APOC packages.
Plugins were copied to the plugins folder (graphaware-nlp-3.3.0.51.1, graphaware-server-enterprise-all-3.3.0.51 and nlp-opennlp-3.3.0.51.1).
The configuration setting were added to neo4j.conf file as given in https://github.com/graphaware/neo4j-nlp
When I restart the Neo4j server, it takes a lot of time and then gives me the following error mesasge:
Caused by: java.lang.OutOfMemoryError: Java heap space Exception in
thread "GraphAware Starter" java.lang.RuntimeException: Error while
initializing model of class: class
opennlp.tools.namefind.TokenNameFinderModel
at
com.graphaware.nlp.processor.opennlp.OpenNLPPipeline.loadModel(OpenNLPPipeline.java:504)
at
com.graphaware.nlp.processor.opennlp.OpenNLPPipeline.lambda$loadNamedEntitiesFinders$2(OpenNLPPipeline.java:162)
at java.util.HashMap$EntrySpliterator.forEachRemaining(Unknown
Without the plugins of NLP, Neo4j starts fine. Any help appreciated here on the minimum requirements of RAM/hardware.

The language models require memory to be loaded. So I would suggest to use at least 4GB or more. Furthermore you should use Enterprise version not CE.

Related

Neo4J Community 4.1.6 Stops after starting

I try to run Neo4J Community Edition 4.1.6 and need a hint. I have downloaded the winzip-package and unzipped it on a Windows 10 PC. I created a folder for the Neo4j files and another for the current JDK 15 package. The path for JDK has been put into system path variables. The log-flags in the neo4j conf has been set to DEBUG for HTPP. JAVA version throws the correct "15.0.2". No JAVA error message while starting.
I can start the service and the status gives back "Neo4j Service started...". After roughly 10 seconds the status shows that Neo4j has stopped without any further notice.
I can't see any logs in the \logs folder - the folder is empty.
I do not see a graphdb been created under neo4j-community-4.1.6\data\databases - the folder is empty
I cannot access the DB by localhost:7474
I do not see any error messages in the cmd windows.
I restarted the system several times but I still have the same outcome. Any hints?
Thank you
I believe we only support JDK 11 when using Neo4j 4.x.
https://neo4j.com/docs/operations-manual/current/installation/requirements/#deployment-requirements-java
Your errors may be because you are using JDK 15.
When diagnosing issues, it also helps to check the neo4j.log and debug.log files in the logs directory.

How to upgrade from Neo4J 2.0.4 to 3.3.3?

I'm trying to upgrade from Neo4J 2.0.4 to 3.3.3 and I followed all the procedures as outlined on the single-instance upgrade page:
http://neo4j.com/docs/operations-manual/3.3/upgrade/deployment-upgrading/
I imported all the configs using the 2.x-config-migrator.jar utility and then ran neo4j-admin import to import the database from 2.0.4 into the 3.3.3, which seems to have worked.
However, when I try to start neo4j with the new database (running neo4j console to see errors as the normal neo4j start didn't work) I see the following error message after this:
2018-02-18 22:59:53.328+0000 INFO Starting upgrade of database
2018-02-18 22:59:53.335+0000 ERROR Failed to start Neo4j: Starting Neo4j failed: Component 'org.neo4j.server.database.LifecycleManagingDatabase#4ae9cfc1' was successfully initialized, but failed to start. Please see the attached cause exception "Not possible to upgrade a store with version 'v0.A.1' to current store versionv0.A.8(Neo4j 3.3.3).
I tried researching for this problem, but there's absolutely no info. The upgrade path 2.0.x(latest) to 3.3.2 is supported, so I guess it's the same for 3.3.3 as it's only a minor upgrade.
When I remove the imported database graph.db from the data folder and just try to run neo4j console on a freshly installed database I also get this error:
2018-02-18 23:09:31.440+0000 ERROR The ResourceConfig instance does not contain any root resource classes.
2018-02-18 23:09:31.440+0000 WARN unavailable The ResourceConfig instance does not contain any root resource classes.
However, when I install a clean database through Neo4J Desktop it runs fine.
I'm using Mac OS X 10.12.6 and Java 1.8.
How could I try to resolve this problem?
In the end, I have to answer this question myself as I spent several hours sorting this out.
First of all, Neo4J claims they support the upgrade path from the latest 2.x release to 3.x but this is not true. I had the version 2.0.4 (which is the latest 2.0 release and there were numerous problems migrating the settings and the old database into the new structure. It would simply not start.
I had to first upgrade to the latest stable 2.x version of Neo4J, which is 2.3.8. I ran into some problems with settings (I used authentification in 2.0.4 but I had to use the newly built in authentication module that 2.3.x provides, so I had to change some settings and remove the auth plugin). You also need to upgrade the database storage from 2.0.x to 2.3.x (by uncommenting the setting in conf).
Once you install and check that 2.3.8 is running and that your old database works, you need to then download the latest 3.x release.
However, the settings migration tool doesn't work properly. It creates numerous settings which are not supported by the new version, so you have to move them manually or not move them at all. Besides, the migration tool doesn't run under Java 1.7 so you might have to install Java 1.8 for the migration tool to even launch... So don't use it.
After that's done, provided you already upgraded your DB to the latest 2.x release, you can run the neo4j-admin import tool to import that DB into the new 3.x version and to convert it.
The whole process is described in detail here: https://github.com/noduslabs/infranodus/wiki/Upgrading-Your-Neo4J-Database-from-2.x-to-3.x

Neo4j: JVM tuning properties ignored

I installed Neo4j Community Edition 2.3.3 on a macbook air with 8GB of RAM. I'm trying to connect to a local database using the web interface.
My local db is quite large: ~200GB, ~600,000,000 nodes and ~1,000,000,000 of relationships.
Using the GUI I went to the options and I edited the file:
/Users/<user>/Documents/Neo4j/.neo4j-community.vmoptions
adding these two lines:
-server
-Xmx4096M
as explained here: http://neo4j.com/docs/stable/performance-guide.html
but those parameters are ignored.
When I try to start the db I get:
Starting Neo4j failed: Component 'org.neo4j.server.database.LifecycleManagingDatabase#b14302e' was successfully initialized, but failed to start. Please see attached cause exception.
And the messages.log file shows:
JVM memory information:
Free memory: 74.67 MB
Total memory: 123.00 MB
Max memory: 1.78 GB
P.S.:
The database was create on a linux (more powerful) machine where everything works fine. I copied the data/ directory locally.
EDIT
I tried with a fresh database, and the problem persists. It looks like neo4j 2.3.3 for mac is just broken. The server runs but the jvm parameters in
.neo4j-community.vmoptions
are ignored. I also tried to write random characters in that file and the JVM runs smoothly. I opened up an issue on github: https://github.com/neo4j/neo4j/issues/6810
UPDATE
The issue has been tagged as an operability bug. No clear solution yet.

After starting hadoop, i cannot start mahout ..!

When Starting Mahout i got an error message like this
root#fazil-VPCEB45FG:/usr/local/mahout/bin# ./mahout
hadoop binary is not in PATH,HADOOP_HOME/bin,HADOOP_PREFIX/bin, running locally
Error occurred during initialization of VM
Could not reserve enough space for object heap
Could not create the Java virtual machine.
i have installed openjdk0.6, Whether openjdk is supported by mahout
There's no OpenJDK 0.6 -- you mean 1.6. Yes Java 6 is supported, as you see in the project documentation. This does not seem to have anything to do with Mahout as it's an error from the JVM itself. The error itself says the problem: you requested a heap that's too large. So, I'd go see what heap you requested in Hadoop config and check it. This is the kind of info you should post in a question.
It's exactly what is says in the error message
Could not reserve enough space for object heap
check your your hadoop config files: hadoop-env.sh and
mapred-site.xml for any properties where you have allocated memory to JVM through the Xmx parameter and lower the values if you don't have enough physical memory.
If you have plenty of ram and you run java on 64-bit OS you need to add a -d64 java option to enforce 64-bit mode (it's not done by default in some cases)
Edit: for stand alone mode (your case) just use a proper Xmx value and -d64 if it is 64bits OS
JAVA_HEAP_MAX parameter in mahout file you're running should be lowered. It was 3GB in the mahout version I downloaded.

CE nightly build does not work out of the box

We have recently installed Orbeon Forms stable 3.8.0 CE's orbeon.war and that works out of box.
Because some features do not work in the stable version we installed the CE nightly build orbeon-CE.war, but this one does not work out of the box.
Form builder has some exceptions. Not only the builder fails but the examples too, so Form runner too.
Java exception
java.lang.NullPointerException
class: java.util.zip.Deflater
method: ensureOpen
line: 421
We are using Linux Debian Lenny 2.6.26-smp, Tomcat 5.5.
Probably we are running into some undocumented requirement for this nightly build...?
Problem identified and workaround:
We use IBM java 1.5 and this is what we found.
Orbeon stable works OK with IBM java 1.5, Orbeon nightly does not work with IBM java 1.5, see previously mentioned zip error.
We then installed SUN java 1.5 and Orbeon nightly works OK with SUN java 1.5., but first there was a problem:
*Exception in thread "exist_QuartzScheduler_QuartzSchedulerThread" java.lang.OutOfMemoryError: PermGen space*
And we googled up this thread:
http://orbeon-forms-ops-users.24843.n4.nabble.com/Data-lost-on-quot-Save-Document-quot-td40450.html
The permgen space is a separate VM
setting. It can be increased with a
VM option, e.g.:
-XX:MaxPermSize=128m
Which solved the problem.
This is not a known issue, and this is something I am unable to reproduce here. Maybe the download you got was corrupted. Try to unzip the war file by hand (e.g. with unzip), and if this gives you an error, re-download a new war. If this doesn't solve your problem, could you update your question with more information, in particular the full stack trace for that NPE?

Resources