Thanks. I opened the PR here: https://github.com/JanusGraph/janusgraph/pull/2760

I confirm, adding lucene-backward-codecs-8.9.0.jar in lib folder solves my problem.

Toom.

Hi, we upgraded Lucene to 8.9.0 version. Do you think the problem will be resolved if we include `lucene-backward-codecs.jar` in the classpath?

Hi,

I'm testing the new pre-release of Janusgraph (from https://github.com/JanusGraph/janusgraph/releases/download/v0.6.0/janusgraph-0.6.0.zip) with the Berkeley/Lucene database created from JG 0.5.3 and all calls of index fail (below the full stacktrace).
Is there a migration process, maybe a reindex ? The error message contains "Could not load codec 'Lucene70'.  Did you forget to add lucene-backward-codecs.jar?". Do you plan to include this jar in your JanusGraph distribution ?

Hi Laura,

One code example says more than 1000 words:

gremlin> graph = TinkerFactory.createModern()
==>tinkergraph[vertices:6 edges:6]
gremlin> g=graph.traversal(
traversal(    traversal()   
gremlin> g=graph.traversal()
==>graphtraversalsource[tinkergraph[vertices:6 edges:6], standard]
gremlin> g.addV().property("lang", 45)
==>v[13]
gremlin> g.V().elementMap()
==>[id:1,label:person,name:marko,age:29]
==>[id:2,label:person,name:vadas,age:27]
==>[id:3,label:software,name:lop,lang:java]
==>[id:4,label:person,name:josh,age:32]
==>[id:5,label:software,name:ripple,lang:java]
==>[id:6,label:person,name:peter,age:35]
==>[id:13,label:vertex,lang:45]
gremlin> g.V().values("lang")
==>java
==>java
==>45
gremlin> g.V().values("lang").group().by(map{it->it.get().getClass()}).by(count())
==>[class java.lang.String:2,class java.lang.Integer:1]
gremlin>

So, this query shows you all occurring data types of a specific property in the graph.
Strictly speaking, gremlin OLAP queries are queries using withComputer(). I tend to use the term a bit looser including analytical queries requiring a full table scan.

Best wishes,

Marc

your graph will be more difficult to use if you do not know what data type is in a property. Users would have to explore for themselves what object types are in the graph. This means a lot of OLAP queries with lots of waiting time and a large load on the graph system.

What would be an example of a OLAP query? There is a way to get a property's type in a gremlin query?

Hi Laura,

Some remarks:

• primitive types can be stored more efficiently than general objects (an integer is exactly 32 bits, an object an be any size)
• for the CompositeIndex the objects are fine as long as they implement the equals() method
• your graph will be more difficult to use if you do not know what data type is in a property. Users would have to explore for themselves what object types are in the graph. This means a lot of OLAP queries with lots of waiting time and a large load on the graph system. If users only use their own data, you can give each user its own graph.
• if you use Gremlin Server, the supported protocols will not know how to serialize arbitrary objects, so you have to instruct the client to request the string values of objects only. Then, you could have asked your users to enter object strings into janusgraph in the first place.

Best wishes,    Marc

ERRATA

The composite index does not work when searching by comparison, ie. .has('name', lt(50)) (I get the usual warning "Query requires iterating over all vertices" and it returns zero vertexes)

I get the warning but the vertex with name=42 *is* returned correctly

I'd like to understand a little bit more about what's going on under the hood when creating a new property with .dataType(Object.class) vs any other specific type eg. .dataType(String.class) or .dataType(Integer.class)
I'm able to create a "name" property like this:

mgmt.makePropertyKey('name').dataType(Object.class).make()

and this is what I've noticed:

- it allows me to create these vertexes
g.addV('alice').property('name', 'Alice')
g.addV('terminator').property('name', 42)
- it allows me to create a composite index, but not a mixed index (confirming what was said in the other thread)
- the composite index works when searching for an exact match, ie. .has('name', 'Alice') and .has('name', 42). The composite index does not work when searching by comparison, ie. .has('name', lt(50)) (I get the usual warning "Query requires iterating over all vertices" and it returns zero vertexes)

I'm only interested into this because I have a graph where multiple people contribute, it would be very nice to not having to deal with explicit property types, if Object.class is an option. For my particular use case I could live without mixed indexes, and I wouldn't mind a small performance deficit (size and/or speed) introduced by the usage of Object as a general type. But I really struggle to understand what's going on. What's the difference between Object and specific types from Janus' point of view? Are types only useful for enforcing a particular schema when inserting data, or there's more to it?



Sent: Wednesday, July 21, 2021 at 10:34 AM
From: hadoopmarc@...
To: janusgraph-users@...
Subject: Re: [janusgraph-users] What are the implications of using Object.class property type?
Hi Laura,

A similar question was posed recently:
https://lists.lfaidata.foundation/g/janusgraph-users/message/5986[https://lists.lfaidata.foundation/g/janusgraph-users/message/5986?p=,,,20,0,0,0::recentpostdate%252Fsticky,,mixedindex,20,2,0,83929827]

So,
1. Only for the CompositeIndex
2. In your specific example, you could use the java Integer class ( https://docs.oracle.com/javase/8/docs/api/java/lang/Integer.html[https://docs.oracle.com/javase/8/docs/api/java/lang/Integer.html] ), because its constructor takes either integer type or string type and it has the equals() method implemented.

Best wishes,    Marc

Hi!

As far as I know, your assumption is correct. However, Janusgraph-GRPC is still in the early stages of development and I guess it will still take a few versions before it will become available. That's also the reason, why the feature is still undocumented. If you have any specific questsions, I would recommend to reach out to Jan Jansen (@farodin91 on GitHub) because he drives the development behind Janusgraph-GRPC.

Best regards,
Florian

Hello!
I've seen some references for JanusGraph-GRPC as something that might replace the ManagementSystem, but can't quite find any more documentation.
Can anyone shed some light?

Many thanks!

Hi Vinayak,

Good to see some progress!

Some suggestions:

• Is 40% relative to a single core or to all cores (e.g. CPU usage for a java process in top can be 800% if 8 cores are present)?
• Ncore * 100% is not necessarily the maximum CPU load of the groovy process + storage backend if the loading becomes IO limited. Can you find out what IO usage is?
• Do you use CompositeIndices on the properties "name" and "e-mail" for the has() filters?
• Regarding the idea from Nicolas, I would rather use a ConcurrentMap that maps ORG id's to vertex id's, but only fill it as you go for the ORG's that you add or lookup. The JanusGraph transaction and database caches should be large enough to hold the vertices to be referenced two or more times, thus accommodating g.V(id) lookups.
• On a single system Apache Spark will not help you.

Best wishes,    Marc

--



Ce message et ses pièces jointes peuvent contenir des informations confidentielles ou privilégiées et ne doivent donc pas être diffusés, exploités ou copiés sans autorisation. 

Hi Vinayak,

Yes, it should be possible to improve on the 3% CPU usage.

The newTransaction() should be added to line 39 (GraphTraversalSource g = graph.traversal();) as the global g from line 121 is not used.

Marc

Hi Vinayak,

What storage backend do you use? Do I understand right that the storage backend and the load script all run on the same server? If, so, are all available CPU resources actively used during batch loading? What is CPU usage of the groovy process and what of the storage backend?

Specific details in the script:

• did you specify storage.batch-loading=true

• I am not sure whether each traversal() call on the graph gets its own thread-independent transaction (that is why ask for the groovy CPU usage). Maybe you need g = graph.newTransaction().traversal() in CsvImporter
• I assume that the global GraphTraversalSource g at the bottom of the script is not used for the bulk loading.

Best wishes,    Marc

Is it expected that tx.log-tx works only for fresh JanusGraph setup?

No. It should work well for your existing JanusGraph setup too. Note that it is a GLOBAL option so it must be changed for the entire cluster. See https://docs.janusgraph.org/basics/configuration/#global-configuration

Best,
Boxuan

Hi All,

I am trying to enable write-ahead logs by using the config property: tx.log-tx to handle transaction recovery.
It's working fine for new JanusGraph setup but not working for Janusgraph setup which already used for some time.

Is it expected that tx.log-tx works only for fresh JanusGraph setup?
Do we need to follow any additional steps to enable tx.log-tx for existing Janusgraph setup.

Thanks,
Radhika