ethereum-optimism/op-geth v1.101411.7

v1.101411.7 - upgrade path to v1.101500.0

on GitHub

This release contains a fix to the upstream journaling code (b79c44a) to ensure no critical state is lost when upgrading full nodes and the journal is thrown away due to a journal format update. Any full node operator who intends to upgrade their v1.101411.x op-geth node to v1.101500.0 should first restart their node once with this release. When you run this release and shut it down again, search the shutdown logs for a message

Persisting dirty state to disk

and its layer argument. It should be lower than 128. If that's the case, you can safely go on to upgrade to v1.101500.0.

As steps:

• You're running v1.101411.x
• Start up v1.101411.7 once, let it run for a while, shut down.
• Check shutdown logs for Persisting dirty state to disk message, check that layer argument is <128
• You're good to upgrade to v1.101500.0

The background is that updating the upstream geth dependencies between v1.101411.x and v1.101500.0, there were two journal format updates in the underlying geth versions from v1.14.11 to v1.15.0. This causes the journal do be dropped at startup of the new release v1.101500.0. A full node should be able to recover from this, but we suspect a bug in geth (ethereum/go-ethereum#31131) means that more than 128 layers are part of the journal, so when it's thrown away at startup, the node may be missing critical state and needs to heal its state.

Full Changelog: v1.101411.6...v1.101411.7