MESS

MESS (Modified Exponential Subjective Scoring, ECBP-1100) makes a node resist deep chain reorganizations. It never changes whether a block is valid. It only decides which of two valid chains the node keeps when a competing chain would replace blocks it already accepted. The MESS confirmation calculator shows how much protection it gives.

Check whether MESS is on¶

1
2

$ geth --classic attach --exec 'admin.ecbp1100Status().enabled' <datadir>/geth.ipc
true

true means MESS is on. false means it is off, or paused while the node is out of sync or has fewer than five peers.

Turn MESS on or off¶

1
2

$ geth --classic --mess          # on
$ geth --classic --mess=false    # off

Without either flag, the node uses the setting its release ships with. Give every node you run the same setting: during a deep reorganization, nodes with different settings can follow different chains.

Which setting fits which operator¶

The setting only matters during a deep reorganization, when a competing chain would replace blocks your node already accepted. With MESS on, your node refuses that chain. With MESS off, it follows whichever chain has the most total difficulty, including one an attacker produced.

For a miner, MESS on has a cost: if other nodes accept a deep reorganization, the blocks your node mines on the chain it kept are orphaned.

Flags¶

• If you give both --mess.activate and --mess.deactivate, the later block wins.
• --mess.activate wins over --mess=false. Flags that contradict each other log a warning at startup.
• The older spellings also work: --ecbp1100, --override.ecbp1100.deactivate and --ecbp1100.nodisable.
• Each flag can also be set with an environment variable, such as GETH_MESS=true or GETH_MESS_ACTIVATE=15000000. The older names, such as GETH_ECBP1100, also work.

Config file¶

Each flag has a line under [Eth] in a config file that does the same thing, and geth dumpconfig run with your flags writes them for you:

A node started from the file keeps the setting until you remove the line. For example, a config file dumped with --mess keeps MESS on until you remove its OverrideECBP1100Deactivate line. A flag given on the command line wins over the file.

Turn MESS on without a restart¶

1
2

$ geth --classic attach --exec 'admin.ecbp1100("latest")' <datadir>/geth.ipc
true

This turns MESS on from the current block. It uses the admin namespace, which the node serves over IPC. At the next start, your flags and config file decide again.

When the node switches it off by itself¶

MESS pauses while the node has fewer than five peers, or while its newest block is more than 390 seconds old. It resumes once the node is back in sync with enough peers. The log shows Disabled artificial finality features with the reason, and Troubleshooting explains it. --mess.nodisable keeps MESS on through both.

The two defaults¶

A release ships one of two defaults, from two Best Practice documents:

Both chains are past the ECBP-1110 blocks, so under ECBP-1110’s default MESS is off, and under ECBP-1100’s it is on. --mess and --mess=false override either one.