[GitHub] [pulsar] bardock opened a new issue #4599: Errors with empty ledgers

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

bardock opened a new issue #4599: Errors with empty ledgers
URL: https://github.com/apache/pulsar/issues/4599
 
 
   In some escenarios, when a broker instance crashes, we find IN_RECOVERY ledgers like this one:
   
   ```
   ledgerID: 00000000-0000-0000-0000-00000102bde6
   BookieMetadataFormatVersion	2
   quorumSize: 3
   ensembleSize: 3
   length: 0
   lastEntryId: -1
   state: IN_RECOVERY
   segment {
     ensembleMember: "10.64.102.142:3181"
     ensembleMember: "10.64.103.19:3181"
     ensembleMember: "10.64.103.132:3181"
     firstEntryId: 0
   }
   digestType: CRC32
   password: ""
   ackQuorumSize: 3
   ```
   
   Causing errors like these:
   
   ```
   [level:ERROR] [class:ManagedLedgerImpl$1] [line:261] [thread:bookkeeper-ml-workers-33-1] - Failed to open ledger 16956902: Error while recovering ledger
   ```
   
   ```
   [level:ERROR] [class:ReadLastConfirmedOp] [line:127] [thread:BookKeeperClientWorker-20-1] - While readLastConfirmed ledger: 16956902 did not hear success responses from all quorums
   ```
   
   ```
   [level:ERROR] [class:PendingAddOp] [line:251] [thread:BookKeeperClientWorker-21-1] - Write of ledger entry to quorum failed: L41015459 E3
   ```
   
   Also, when empty ledgers are marked as under-replicated, they slow down the entire replication process. 
   
   We found that manually deleting these empty ledgers fixes both problems.
   
   Do you know if a newer version of pulsar fix this issue?
   
   #### System configuration
   **Pulsar version (brokers)**: 1.20.0
   **Bookkeeper version**: 4.7.2

----------------------------------------------------------------
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:
users@infra.apache.org


With regards,
Apache Git Services