Uploaded image for project: 'Fabric'
  1. Fabric
  2. FAB-18285

The new peer node would not update ledger if the channel committed a chaincode with 'init-required=false' flag

    XMLWordPrintable

Details

    • Bug
    • Status: Closed
    • High
    • Resolution: Invalid: Environment Issue
    • v2.2.0
    • None
    • fabric-peer
    • Hide
      Step1：
      use "init-required=false" flag to approve the chaincode definition on peer0 node of each org, then commit it.
      Step2：
      Enter the peer1 container (of any org), fetch the zero block and join the target channel.
      Step3:
      Then this bug will have a great possibility to happend. When you use 'peer channel getinfo -c mychannel' command to check the block status,
      you would find that the blockchain height stay at 1 and never change.
      Sometimes——i would rather say——seldom，it may not happen，but when you try some more times，it would bother you too.
      Show
      Step1： use "init-required=false" flag to approve the chaincode definition on peer0 node of each org, then commit it. Step2： Enter the peer1 container (of any org), fetch the zero block and join the target channel. Step3: Then this bug will have a great possibility to happend. When you use 'peer channel getinfo -c mychannel' command to check the block status, you would find that the blockchain height stay at 1 and never change. Sometimes——i would rather say——seldom，it may not happen，but when you try some more times，it would bother you too.

    Description

      Description

      I found a bug while employing a chaincode with 'init-required=false' flag on the peer node.
      When this bug happens, the new joining node can't get any ledger through gossip protocol.
      I have no idea which parts of code work wrong, but i spend some time to ensure how to reproduce the bug using fewest steps.
      i think it only happens when you join a new peer (like peer1) of an existing org (peer0 has joined) of the channel.

      Condition

      Fabric-network: 1 consortium with 3 orgs, each org has two peers, named after peer0 and peer1. （maybe not necessary，but mention it in case）
      configuration:

      • CORE_PEER_GOSSIP_USELEADERELECTION=true
      • CORE_PEER_GOSSIP_ORGLEADER=false

      Workaround

      i found a workaround to solve this problem.
      you can stop the peer0 docker，wait for a minute，then restart it.
      now you will see the peer1 start to fetch ledger and work well.
      i guess its because peer1 becomes the leader peer of this org when the peer0 down.
      but I cannot analyze it from the code level
      i also try to restart peer1 only, well, it just dosent work.

      Attachments

        Activity

          People

            denyeart David Enyeart
            sennri Yuqi Ni
            Votes:
            0 Vote for this issue
            Watchers:
            3 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: