Windows users guide¶
This guide upgrades a node on Windows from any v1.12.x release to v1.13.0. Migrating to v1.13.0 explains why, and what changes for every node.
The commands are for PowerShell. They use the default Ethereum Classic data directory, %LOCALAPPDATA%\Ethereum\classic. The node uses %USERPROFILE%\AppData\Roaming\Ethereum\classic instead if an older AppData\Roaming\Ethereum folder exists and is not empty. If you pass --datadir, use yours; for Mordor, use mordor in place of classic.
On Windows the node listens for geth attach on a named pipe, \\.\pipe\geth.ipc, not on a file in the data directory.
Choose how you update¶
| Method | Use it if |
|---|---|
| Release archive | You run the published binary, core-geth-win64 |
| Build from source | You compile the client yourself |
Before you start¶
Record what you are running, from the folder that holds geth.exe:
1 2 3 4 | |
Keep that last line. It is your current enode ID, and step 3 changes it.
1. Stop the node cleanly¶
- In a console window: press Ctrl+C once and wait.
- As a service: stop it with the tool you installed it with, and give it time to shut down cleanly.
Wait for INFO [..] Blockchain stopped before you continue. It confirms the database is consistent.
2. Install v1.13.0¶
Keep the old binary until the upgrade is verified; step 6 uses it.
Method A: release archive¶
Download and check core-geth-win64-v1.13.0.zip as Installation, Windows shows, then verify where it came from. In the folder that holds your current geth.exe:
1 2 3 | |
Method B: build from source¶
Building needs Go 1.26 and a C compiler on PATH, such as MinGW-w64. Build from the release branch, not from main:
1 2 3 4 | |
That is the command make geth runs on other systems. Rename your current geth.exe to geth-v1.12.exe, then copy build\bin\geth.exe in its place. Do not clone with --recursive: the submodules hold test fixtures of several gigabytes that the build does not read.
3. Rotate the P2P node key¶
Required. Rotate the P2P node key explains why, and which peer lists to update. With the node stopped:
1 2 | |
4. Re-check RPC exposure¶
Go through the RPC exposure checklist against the flags you start the node with.
5. Start and verify¶
Start the node the way you always do. Then, in this order:
1 2 3 4 | |
A peer count stuck at zero after ten minutes usually means a peer still lists your old enode ID. Last, compare eth.blockNumber with a block explorer or another node you operate.
6. If you need to roll back¶
The data directory works in both directions. Stop the node, then restore the old binary and start it again:
1 2 | |
The rotated node key stays rotated: rolling back does not un-expose the old key.