Skip to content
Last update: January 17, 2022

Upgrade GoQuroum

The following information is for upgrading GoQuorum only.

You can have multiple versions of GoQuorum running on the same network, allowing you to perform a rolling (node-by-node) upgrade. Ensure your network maintains full uptime availability when upgrading to provide redundancy and reversibility if anything goes wrong.

Incremental upgrades

Stop the GoQuorum node you wish to upgrade, update the binary to the next version, and restart the node. Check that the node starts without errors and has the latest block from the network.

We don’t recommend jumping versions during an upgrade; some versions require manual intervention. Check the release log for any actions you might need to take.

Upgrade to GoQuorum 2.6.0

GoQuorum 2.6.0 upgrades the base Geth version from 1.8.18 to 1.9.7. See Ethereum 1.9.0 for the complete list if new features added as a part of Geth 1.9.7.

Note

Geth 1.9.7 has several enhancements at the database layer which are part of GoQuorum 2.6.0. Once you migrate to 2.6.0, it can’t roll back to an older version of GoQuorum. We recommend backing up the data directory before upgrading to 2.6.0, which you can use to revert back to the older version if necessary.

A node running on GoQuorum 2.6.0 can coexist in a network with other nodes on lower versions of GoQuorum, so you can perform a rolling upgrade to GoQuorum 2.6.0. The following is the recommended upgrade process:

  1. Stop the node you wish to upgrade to GoQuorum 2.6.0. Modify the genesis.json file to include istanbulBlock and petersburgBlock. Set these parameters to an appropriate value in future by when the entire network will be upgraded to GoQuorum 2.6.0.

    Warning

    Setting the milestone blocks is necessary because the gas calculation logic changes in Geth 1.9.7. Not setting these values properly can result in a Bad block error.

  2. GoQuorum 2.6.0 deprecates genesis parameters maxCodeSize and maxCodeSizeChangeBlock. A new attribute maxCodeSizeConfig is added to genesis to allow tracking of multiple maxCodeSize value changes.

    In earlier GoQuorum versions, if the maxCodeSize is changed multiple times, any node joining the network might get a Bad block error. The changes in GoQuorum 2.6.0 enable tracking of historical changes of maxCodeSize in the genesis file and thus allow it to be changed multiple times in the network’s life. When init is executed in GoQuorum 2.6.0, Geth forces use of maxCodeSizeConfig.

    Example

    "config": {
    "homesteadBlock": 0,
    "byzantiumBlock": 0,
    "constantinopleBlock": 0,
    "petersburgBlock": 0,
    "istanbulBlock": 0,
    "chainId": 10,
    "eip150Block": 0,
    "eip155Block": 0,
    "eip150Hash": "0x0000000000000000000000000000000000000000000000000000000000000000",
    "eip158Block": 0,
    "isQuorum": true,
    "maxCodeSizeConfig": [
      {
        "block": 5,
        "size": 35
      },
      {
        "block": 15,
        "size": 24
      },
      {
        "block": 20,
        "size": 35
      }
    ]
  },

  3. Execute geth --datadir path/to/datadir init genesis.json with the modified genesis.json.

  4. Start the node in GoQuorum 2.6.0.

Notes

  • Geth 1.9.7 introduces freezerdb, where block data beyond a certain threshold is moved to a different file-based storage area. Provide the location for freezerdb using the --datadir.ancient Geth command line option.

  • When a node is migrated to this version, Geth by default creates the ancient data folder and starts moving blocks below the immutability threshold (default: 3162240) into the ancient data.

    If you don’t want this movement to happen, use --immutabilitythreshold to set the immutability threshold to an appropriate value when starting Geth.

  • Geth 1.9.7 by default doesn’t allow unlocking keystore-based accounts in the startup process, and crashes if you attempt this. To enable account unlocking, use the --allow-insecure-unlock Geth option.

  • In a GoQuorum 2.6.0 network running with gcmode=full and block height exceeding the immutability threshold (with blocks in freezerdb), if a node is restarted non-gracefully (kill -9/docker kill & start), it can fail to sync up with its peers, generating a missing parent error. This is due to an upstream bug where non-graceful restart causes a gap between leveldb and freezerdb.

    You can avoid this by either running the node with gcmode=archive or restarting the node gracefully (kill / docker stop & start).

    This is fixed in GoQuorum v21.4.0 (from upstream Geth 1.9.20).

Back to top