https://linen.dev logo
Join Slack
Powered by
This message was deleted.
# troubleshooting
s
This message was deleted.
s
After the restart, are the supervisors in Druid still running? Are they healthy? You can see this in the Druid Console in the Ingestion View.
i
Yes, it's healthy and running
Copy code
{
  "dataSource": "",
  "stream": "",
  "partitions": 1,
  "replicas": 1,
  "durationSeconds": 3600,
  "activeTasks": [
    {
      "id": "",
      "startingOffsets": {
        "0": 11305
      },
      "startTime": "2022-11-22T16:14:00.921Z",
      "remainingSeconds": 1304,
      "type": "ACTIVE",
      "currentOffsets": {
        "0": 11305
      },
      "lag": {
        "0": -9753
      }
    }
  ],
  "publishingTasks": [],
  "latestOffsets": {
    "0": 1552
  },
  "minimumLag": {
    "0": -9753
  },
  "aggregateLag": 0,
  "offsetsLastUpdated": "2022-11-22T16:52:06.402Z",
  "suspended": false,
  "healthy": true,
  "state": "RUNNING",
  "detailedState": "RUNNING",
  "recentErrors": []
}
s
Do you see the real-time segments for the datasource in the Segments view? The broker depends on the announcement of segments to recognize the data that is available. Normally (or by default) this happens through Zookeeper. Could there be something wrong with zookeeper after the restart? Another thought is that you can change this to use the HTTP communication instead by setting the 
druid.serverview.type=http
instead of 
batch
for segment discovery in the Broker.
i
I've noticed that when the system restarts, kafka resets the offset. While, druid keeps the old offset.
s
Oh, I see. You can use the "hard reset" from the supervisor's Actions menu to reset the offsets. These are the relevant docs.
Open in Slack
PreviousNext
Powered by