Hey there! Let's break down this odd issue you're hitting with JanusGraph—specifically why certain property keys like abc and Mailing_Code throw type errors when inserting float values, while others like a work just fine.

Most Likely Cause: Predefined Property Key Types #

JanusGraph is a strongly typed graph database, which means every property key has an explicit data type assigned (either by you upfront, or auto-inferred when you first use the key).

The error message Value [9.5] is not an instance of the expected data type for property key [abc] and cannot be converted. Expected: class java.lang.Integer, found: class java.lang.Float tells us exactly what's going on:

• You (or JanusGraph's auto-inference) previously set the abc and Mailing_Code keys to expect Integer values
• When you try to insert a Float, the database rejects it because it doesn't match the predefined type
• For the a key, since it was never defined before, JanusGraph auto-inferred it as a numeric type that supports floats, hence no error.

How to Fix This #

Step 1: Verify the Property Key's Type #

First, confirm the data type of the problematic keys by running these Gremlin commands:

# Check the type for "abc"
g.propertyKey("abc").valueMap()

# Check the type for "Mailing_Code"
g.propertyKey("Mailing_Code").valueMap()

Look for the dataType field in the output—if it shows class java.lang.Integer, that's the root cause.

Step 2: Resolve the Type Mismatch #

Depending on whether you already have data associated with these keys, choose one of these solutions:

Option A: No Existing Data (Fresh Setup)

If you haven't stored any critical data yet, delete the misconfigured property keys and recreate them with the correct Float type:

# Drop the old Integer-type keys
g.propertyKey("abc").drop()
g.propertyKey("Mailing_Code").drop()

# Recreate them with Float data type
g.propertyKey("abc").dataType(Float.class).create()
g.propertyKey("Mailing_Code").dataType(Float.class).create()

After this, your g.addV("T22").property("abc", 9.5f) command should work as expected.

Option B: Existing Data (Avoid Data Loss)

If you already have data using these keys, you can't directly modify the existing property key's type. Instead:

1. Create a new property key with the correct Float type (e.g., abc_float or Mailing_Code_float)
2. Migrate existing data from the old key to the new one
3. Update your queries to use the new key going forward
4. (Optional) Once all data is migrated, drop the old Integer-type key

Pro Tip for JanusGraph Newbies #

Instead of relying on JanusGraph's auto-inference for property keys, explicitly define all your property keys and their data types upfront. This eliminates unexpected type mismatches like this one. For example:

# Define all required property keys with explicit types
g.propertyKey("abc").dataType(Float.class).create()
g.propertyKey("Mailing_Code").dataType(Float.class).create()
g.propertyKey("a").dataType(Float.class).create()

内容的提问来源于stack exchange，提问作者Akshay