Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 14 additions & 0 deletions website/docs/engine-flink/lookups.md
Original file line number Diff line number Diff line change
Expand Up @@ -289,6 +289,20 @@ partitioning enabled and exactly one partition key. When enabled, the Coordinato
retains the `__historical__` system partition used to route lookups to Paimon. Disabling the option
removes that system partition.

Historical partition lookups run on TabletServers. Configure `datalake.format` and the related
`datalake.<format>.*` options on every TabletServer, for example:

```yaml title="server.yaml"
datalake.enabled: true
datalake.format: paimon
datalake.paimon.metastore: filesystem
datalake.paimon.warehouse: /path/to/paimon/warehouse
```

Also make the Paimon connector, `paimon-bundle`, and required catalog or storage JARs available on

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Just curious about do we really need to mentention about it since it's already in fluss distribution. But It reminds me that we did need to put paimon oss jar into ${FLUSS_HOME}/plugins/paimon/.

Also, i feel like this page may not a right place to put this. Maybe https://fluss.apache.org/docs/next/install-deploy/deploying-streaming-lakehouse/ or https://fluss.apache.org/docs/next/streaming-lakehouse/datalake-formats/paimon/. But we can add a link in this page for how to make historical partition lookups works

every TabletServer. Use `${FLUSS_HOME}/plugins/paimon/` for the binary distribution or
`${FLUSS_HOME}/lib/` for a flat classpath, then restart the TabletServers after changes.

Lookup clients use the table configuration captured when the lookuper is created to decide whether
to fall back after an original partition is missing. After changing
`table.datalake.historical-partition.enabled`, restart existing lookup jobs that need to look up
Expand Down