smit-aterlo opened a new issue #10672:
URL: https://github.com/apache/druid/issues/10672


   Druid Supervisor fails with `LOST_CONTACT_WITH_STREAM` after running for a 
while and it happens after 1 or 2 Kafka log clean-up. 
   
   ### Affected Version
   
   0.20.0
   
   ### Description
   
   Please include as much detailed information about the problem as possible.
   
   - Cluster Size: 1 Master, 1 Query (Both 16 CPUs and 64GB Memory), 2 Data 
Nodes (Both 64 CPUs and 196GB Memory)
   - A number of Topics: 5 Topics (1 big topic is with 1000 partitions)
   - Ingestion Rate: 4 Topic Ingest around 100k each and 1 with around 1 million
   - Supervisor Config for biggest topic:
   ```
   {
       "type": "kafka",
       "spec": {
           "ioConfig": {
               "type": "kafka",
               "topic": "table1",
               "consumerProperties": {
                   "bootstrap.servers": "10.128.0.96:9092"
               },
               "inputFormat": {
                   "type": "json"
               },
               "useEarliestOffset": true,
               "appendToExisting": true
           },
           "tuningConfig": {
               "type": "kafka",
               "resetOffsetAutomatically": true
           },
           "dataSchema": {
               "dataSource": "table1",
               "granularitySpec": {
                   "type": "uniform",
                   "queryGranularity": "NONE",
                   "segmentGranularity": "HOUR",
                   "rollup": false
               },
               "timestampSpec": {
                   "column": "Time",
                   "format": "millis"
               },
               "dimensionsSpec": {
                   "dimensions": [
                       "AddressSpace",
                       "Field1",
                       "Field2",
                       {
                           "type": "long",
                           "name": "Gauge"
                       },
                       {
                           "type": "long",
                           "name": "Gauge2"
                       },
                       "Field3",
                       "Field4",
                       "Field4",
                       "Field5",
                       "Field6",
                       "Field7",
                       "Field8"
                   ]
               }
           }
       }
   }
   ```
   
   - Steps to Reproduce: Druid can consume all the messages until 1st or 2nd 
log clean up runs on the Kafka side. Once it runs Supervisor fails with 
`LOST_CONTACT_WITH_STREAM` and the following error:
   
   ```
   {
         "timestamp": "2020-12-11T12:09:19.517Z",
         "exceptionClass": "org.apache.druid.java.util.common.ISE",
         "message": "org.apache.druid.java.util.common.ISE: Previous 
sequenceNumber [716362] is no longer available for partition [461] - 
automatically resetting sequence",
         "streamException": true
       }
   ```
   
   Can someone please help me with this issue?
   
   Thank you
   


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
[email protected]



---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to