klyashko opened a new issue, #5004: URL: https://github.com/apache/paimon/issues/5004
### Search before asking - [x] I searched in the [issues](https://github.com/apache/paimon/issues) and found nothing similar. ### Paimon version Not sure that it's a bag, but wanted to clarify the question of schema evolution. Currently we are onboarding Paimon in way: 1. There is a Flink application which injects data to Paimon tables 2. There are readers which consumes the data later During experimenting with schema evolution noticed that when schema change was done in external way Flink application didn't pick up changed properties even after it was restarted. For application to pick up the change I had to restart job without its state. What riese a question how to properly manage the schema evolution, should that be imbedded to the application as pre job execution steps or should that be managed somehow else? ### Compute Engine Flink ### Minimal reproduce step 1. Write to Paimon table from stream application in way: ` val table = env.sqlQuery(query) val data = env.toChangelogStream(table) val catalog = FlinkCatalogFactory.createPaimonCatalog(...) val pTable = catalog.getTable(Identifier.create(..., ...)) new FlinkSinkBuilder(pTable) .forRow(data, table.getSchema.toRowDataType) .parallelism(1) .build() ` Usage of `HiveCatalog` didn't change the behaviour. 2. Change Paimon table property with external Java application based on Catalog API documentation section ### What doesn't meet your expectations? Absence of understanding after consulting with documentation how safely evolve table schema. In our case it was attempt to change `snapshot.time-retained`. And preferably do it without losing job state ### Anything else? _No response_ ### Are you willing to submit a PR? - [ ] I'm willing to submit a PR! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
