Glue Schema Registry Checkpoint Definition

Question

I have some questions about Glue Schema Registry Checkpoint.

The checkpoint definition in AWS documentation doesn't seem to be right. In this [glue schema documentation](https://docs.aws.amazon.com/glue/latest/dg/schema-registry.html) under `schema versioning and compatibility` section, it says

```
A schema version that is marked as a checkpoint is used to determine the compatibility of registering new versions of a schema.
```

However, I find that is not true. Based on my experiment, when you register a new version of schema, it always need to be compatible with previous version. For example,
* v3 needs to be compatible with v2,
* v2 needs to be compatible with v1

Based on the documentation, it implies that
* v3 just needs to be compatible with v1 not v2 given that v1 is checkpoint version.

but that will give me a failed status schema saying incompatible schema definition
Here is another bit of [documentation](https://docs.aws.amazon.com/glue/latest/dg/schema-registry-gs.html) in a different page says
```
If the version changed the compatibility mode, the version will be marked as a checkpoint.
```
This is looking right to me, as I found that if user need to change the schema compatibility, they need to change the checkpoint to be the latest version.

To conclude, I wonder
* what is its actual definition?
* what is it used for?
* does this affect schema validation?

Accepted Answer

Hi,

The Glue Schema Registry Checkpoint changes to the most recent version of the schema by default when we change/modify the schema definition or compatibility mode.If you want to specify a specific version of the schema, you must use CLI/SDK; you can specify the version using the updateSchema API, and you can also specify the latest version of the schema.

Where as when a new version of a schema is submitted to the registry, the compatibility rule applied to the schema name is used to determine if the new version can be accepted. There are 8 compatibility modes: NONE, DISABLED, BACKWARD, BACKWARD_ALL, FORWARD, FORWARD_ALL, FULL, FULL_ALL. Here the BACKWARD compatibility implies that the new schema will be compared to the checkpoint schema. If the schemas do not match, then it will reject the new version of the schema. With "backward" compatibility, you are only able to delete fields and add new fields.

Thank you.

Glue Schema Registry Checkpoint Definition

Conteúdo relevante