Hi everyone! I want to replace a drive. I found the migration process, but I have many questions. If my identity and data are on a different drive, am I correct in assuming I don’t need to copy them? I would only be replacing the data drive.
The instructions say to copy the data from the command line first and then stop the service. Why not stop it beforehand? A regular copy takes three days, which seems way too long. If I don’t stop the service, it keeps updating the data during the process, which doesn’t seem like a good solution to me.
Q1) Yes the identity must stay the same. If you have put some folders like identity, logs, etc. on a separate drive, just change the data location before restarting the node.
Q2) It’s recommended to do a first pass copy in the background to avoid extended downtime, then shut down the node and do a final sync pass which should be much quicker and will pickup all the latest changes.
Typically you would use rsync/rclone (linux) or robocopy /mir (windows). You can temporarily set the max size of your node to something lower to avoid ingress activity while doing all that.
4 Likes
Now I understand, thank you for the answer!
3 Likes
After 4 h of downtime, every piece that’s requested from your node will be considered lost and it will be repaired on someone else’s node. When you come back online, your node will receive new bloom filters that will clean those “lost” pieces by sending them to trash.
So, you can stop the ingress by lowering the allocated space to a minimum (500GB), but you can’t stop the egress, unless you are OK with the data loss.
These days, with so low data gains, every bit of stored data is gold. But it’s your choise. Your node will not be DQed with under 30 days offline.
1 Like