Using Flink CLI
You can use the Flink command line interface to operate, configure and maintain your Flink applications.
The Flink CLI works without requiring the user to always specify the YARN application ID when submitting commands to Flink jobs. Instead, the jobs are identified uniquely on the YARN cluster by their job IDs.
The following table summarizes the supported actions of Flink CLI:
| Action | Purpose |
|---|---|
run |
You can use this action to execute jobs. It requires at least the jar containing the job. Flink- or job-related arguments can be passed if necessary. |
run-application |
You can use this action to execute jobs in application mode. It requires the same
parameters as the run action. |
info |
You can use this action to print an optimized execution graph of the passed job. It requires the jar containing the job. |
list |
You can use this action to list all running or scheduled jobs. |
savepoint |
You can use this action to create or dispose savepoints for a given job. You might
need to specify a savepoint directory besides the job ID if the
state.savepoints.dir was not specified in Cloudera Manager. |
checkpoint |
You can use this action to create checkpoints for a given job. The checkpoint type (full or incremental) can be specified. |
cancel |
You can sue this action to cancel running jobs based on their job ID. |
stop |
You can use this action to stop a running job, but also create a savepoint to start
from again. This action combined cancel and
savepoint. |
