Moving node with hashstore to new drive

I moved a 7TB node to a new drive, node already migrated to hashstore, and it took 11 hours. Both drives are on the same PC, Exos SATA drives, OS on NVMe, 32GB RAM, Ubuntu server, ext4 fs. It is way faster than moving a piecestore node. I remember it took me like 1 day/TB.
Here is what I used:

docker ps -a
docker stop -t 300 storagenode
docker rm storagenode
rsync -aP --delete /mnt/hdd1p1/Storj1/ /mnt/hdd2p1/Storj2/

Node is run as user, not as root, and the commands are run also as user.
Both source and destination have the same owner and permissions.
I understand that the --delete flag deletes the source content.
Well, it didn’t… Why? What did I miss?

1 Like

Doesn’t the “–delete” flag erase anything in the destination folder that’s not in the source folder? And if the destination was the new empty drive…

(you’re syncing files: and that flag keeps old stuff from piling up in destinations: like if you’re syncing daily or something)

4 Likes

https://www.cyberciti.biz/faq/linux-unix-bsd-appleosx-rsync-delete-file-after-transfer/

2 Likes

Yeah, my mistake. I didn’t read the manual. I imagined it deletes the source, like the move command.

You meant --remove-source-files

2 Likes

i would just ude “dd” and then gparted to enlarge the partition

of course there will be some downtime, but will recover from that

I needed to clean that drive, because I wanted to change the block size. Wasn’t a matter of space.

Good to see, hashstore is much faster, than piecestore, I am migrating right now, to move my nodes.

2 Likes

:musical_notes: I like big blocks…
… and I cannot lie… :musical_notes:

:man_dancing: :woman_dancing:

2 Likes

I misread “blocks” in the first seconds

1 Like

That is a very encouraging data point, migrating nodes with pieces store was always the worst.