J-HowHuang commented on code in PR #16886:
URL: https://github.com/apache/pinot/pull/16886#discussion_r2383578225
##########
pinot-controller/src/main/java/org/apache/pinot/controller/api/resources/PinotTenantRestletResource.java:
##########
@@ -701,6 +705,46 @@ public SuccessResponse deleteTenant(
Response.Status.INTERNAL_SERVER_ERROR);
}
+ @DELETE
+ @Produces(MediaType.APPLICATION_JSON)
+ @Authenticate(AccessType.DELETE)
+ @Authorize(targetType = TargetType.CLUSTER, action =
Actions.Cluster.REBALANCE_TENANT_TABLES)
+ @Path("/tenants/rebalance/{jobId}")
+ @ApiOperation(value = "Cancels a running tenant rebalance job")
+ @ApiResponses(value = {
+ @ApiResponse(code = 200, message = "Success", response =
SuccessResponse.class),
+ @ApiResponse(code = 404, message = "Tenant rebalance job not found"),
+ @ApiResponse(code = 500, message = "Internal server error during
cancelling the rebalance job")
+ })
+ public SuccessResponse cancelRebalance(
+ @ApiParam(value = "Tenant rebalance job id", required = true)
@PathParam("jobId") String jobId) {
+ Map<String, String> jobMetadata =
+ _pinotHelixResourceManager.getControllerJobZKMetadata(jobId,
ControllerJobTypes.TENANT_REBALANCE);
+ if (jobMetadata == null) {
+ throw new ControllerApplicationException(LOGGER, "Failed to cancel
tenant rebalance job: " + jobId,
+ Response.Status.NOT_FOUND);
+ }
+ try {
+ TenantRebalanceContext originalContext =
TenantRebalanceContext.fromTenantRebalanceJobMetadata(jobMetadata);
+ TenantRebalanceProgressStats progressStats =
+
JsonUtils.stringToObject(jobMetadata.get(RebalanceJobConstants.JOB_METADATA_KEY_REBALANCE_PROGRESS_STATS),
+ TenantRebalanceProgressStats.class);
+ originalContext.getParallelQueue().clear();
+ originalContext.getSequentialQueue().clear();
+ TenantRebalancer.TenantTableRebalanceJobContext ctx;
+ while ((ctx = originalContext.getOngoingJobsQueue().poll()) != null) {
Review Comment:
It's a good point. Then the API will not cancel the newly picked table
rebalance job, there should be no job picked once the ZK update done.
I felt like this implementation is a bit buggy, so I made the job
cancellation logic into the observer. The code should be easier to read and
bug-free for this cancellation api as well as the `TenantRebalanceChecker` now.
Could you take a look?
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
