[GitHub] [druid] kaijianding opened a new pull request #11515: should retry when failed to pause realtime task while task is publishing

[GitBox <gi...@apache.org>]

kaijianding opened a new pull request #11515:
URL: https://github.com/apache/druid/pull/11515


   I noticed that realtime task is killed when `CheckpointNotice` try to pause the task while the task is in `PUBLISHING` status.
   
   in `SeekableStreamIndexTaskRunner`
   ```java
   
     public Response pause() throws InterruptedException
     {
       if (!(status == Status.PAUSED || status == Status.READING)) {
         return Response.status(Response.Status.BAD_REQUEST)
                        .entity(StringUtils.format("Can't pause, task is not in a pausable state (state: [%s])", status))
                        .build();
       }
       ...
     }
   ```
   
   in `IndexTaskClient.submitRequest`
   ```java
         try {
           // Netty throws some annoying exceptions if a connection can't be opened, which happens relatively frequently
           // for tasks that happen to still be starting up, so test the connection first to keep the logs clean.
           checkConnection(request.getUrl().getHost(), request.getUrl().getPort());
   
           response = submitRequest(request);
   
           int responseCode = response.getStatus().getCode();
           if (responseCode / 100 == 2) {
             return response;
           } else if (responseCode == 400) { // don't bother retrying if it's a bad request
             throw new IAE("Received 400 Bad Request with body: %s", response.getContent());
           } else {
             throw new IOE("Received status [%d] and content [%s]", responseCode, response.getContent());
           }
         }
         catch (IOException | ChannelException e) {
             // retry logic here
             ...
         }
   ```
   
   Response.Status.BAD_REQUEST is 400, as a result of the `IAE` exception, the `CheckpointNotice` is marked as failed and the realtime task is asked to shutdown. After that a new task is pulled up to start over from the previous checkpointed offset which is done like one hour ago.
   
   I think it is better to retry this `pause` to wait for the publishing finished and finally the offset checkpoint can success, rather than a force shutdown
   
   <hr>
   
   ##### Key changed/added classes in this PR
    * `SeekableStreamIndexTaskRunner`
   
   <hr>
   
   <!-- Check the items by putting "x" in the brackets for the done things. Not all of these items apply to every PR. Remove the items which are not done or not relevant to the PR. None of the items from the checklist below are strictly necessary, but it would be very helpful if you at least self-review the PR. -->
   
   This PR has:
   - [ x ] been self-reviewed.
      - [ ] using the [concurrency checklist](https://github.com/apache/druid/blob/master/dev/code-review/concurrency.md) (Remove this item if the PR doesn't have any relation to concurrency.)
   - [ ] added documentation for new or modified features or behaviors.
   - [ ] added Javadocs for most classes and all non-trivial methods. Linked related entities via Javadoc links.
   - [ ] added or updated version, license, or notice information in [licenses.yaml](https://github.com/apache/druid/blob/master/dev/license.md)
   - [ ] added comments explaining the "why" and the intent of the code wherever would not be obvious for an unfamiliar reader.
   - [ ] added unit tests or modified existing tests to cover new code paths, ensuring the threshold for [code coverage](https://github.com/apache/druid/blob/master/dev/code-review/code-coverage.md) is met.
   - [ ] added integration tests.
   - [ ] been tested in a test Druid cluster.
   


-- 
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: commits-unsubscribe@druid.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org

[GitHub] [druid] jiaqi-g commented on pull request #11515: should retry when failed to pause realtime task while task is publishing

[GitBox <gi...@apache.org>]

jiaqi-g commented on pull request #11515:
URL: https://github.com/apache/druid/pull/11515#issuecomment-915730614


   Hi @kaijianding Thanks so much for this diff, looks very useful as we have seen many times the druid realtime task is cancelled by overlord when publishing, which is very annoying. I will pull your diff and have a try. Btw, have you taken a look at the re-try logic after CONFLICT status is reported, is it retrying only one time or multiple times? and what is the time interval and how can we configure it? Thanks!


-- 
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: commits-unsubscribe@druid.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org