Behavioral Changes in Schema Registry

Learn about the change in certain functionality of Schema Registry that has resulted in a change in behavior from the previously released version to this version of Cloudera Runtime.

Summary:
SchemaRegistryClient#runRetryableBlock is not thread safe
Previous behavior:
The default backoff policy for the Cloudera distributed Schema Registry Java client was com.hortonworks.registries.schemaregistry.retry.policy.NOOPBackoffPolicy. This backoff policy did not allow multiple retries even if more Schema Registry URLs were in use where multiple retries can occur.
New behavior:
The default backoff policy of the Schema Registry client has changed to com.hortonworks.registries.schemaregistry.retry.policy.ExponentialBackoffPolicy. This backoff policy allows retries of requests with the default configuration. If retries are not desirable in a client application, users need to explicitly set the schema.registry.client.retry.policy.className to com.hortonworks.registries.schemaregistry.retry.policy.NOOPBackoffPolicy.
Summary:
Increased SR server DB timeout for JdbcStorageManager#nextId
Previous behavior:
The Schema Registry database could easily timeout under heavy workload.
New behavior:
The database write timeout was increased from 3 seconds to 10 seconds. Additionally, the timeout related exception mapping is changed to be more informative and is retryable by the client.