Re: btrfs send / receive via netcat, fails halfway?

[Forza <forza@tnonline.net>]

On 2021-01-10 11:34, Cedric.dewijs@eclipso.eu wrote:
> ­I'm trying to transfer a btrfs snapshot via the network.
> 
> First attempt: Both NC programs don't exit after the transfer is complete. When I ctrl-C the sending side, the receiving side exits OK.
> 
> btrfs subvolume delete /mnt/rec/snapshots/*
> receive side:
> # nc -l -p 6790 | btrfs receive /mnt/rec/snapshots
> At subvol 0
> 
> sending side:
> # btrfs send  /mnt/send/snapshots/0 | nc -v 127.0.0.1 6790
> At subvol /mnt/send/snapshots/0
> localhost [127.0.0.1] 6790 (hnmp) open
> 
> 
> 
> Second attempt: both nc programs exit ok at snapshot 0,1,2, but snapshot3 fails halfway, and 4 fails, as 3 is not complete.
> receive side:
> # nc -l -p 6790 | btrfs receive /mnt/rec/snapshots
> At subvol 0
> # nc -l -p 6790 | btrfs receive /mnt/rec/snapshots
> At snapshot 1
> # nc -l -p 6790 | btrfs receive /mnt/rec/snapshots
> At snapshot 2
> # nc -l -p 6790 | btrfs receive /mnt/rec/snapshots
> At snapshot 3
> read(net): Connection reset by peer
> ERROR: short read from stream: expected 49183 read 10450
> # nc -l -p 6790 | btrfs receive /mnt/rec/snapshots
> At snapshot 4
> ERROR: cannot find parent subvolume
> write(stdout): Broken pipe
> 
> sending side:
> # btrfs send  /mnt/send/snapshots/0 | nc -v -c 127.0.0.1 6790
> At subvol /mnt/send/snapshots/0
> localhost [127.0.0.1] 6790 (hnmp) open
> # btrfs send -p /mnt/send/snapshots/0 /mnt/send/snapshots/1 | nc -v -c  127.0.0.1 6790
> At subvol /mnt/send/snapshots/1
> localhost [127.0.0.1] 6790 (hnmp) open
> # btrfs send -p /mnt/send/snapshots/1 /mnt/send/snapshots/2 | nc -v -c  127.0.0.1 6790
> At subvol /mnt/send/snapshots/2
> localhost [127.0.0.1] 6790 (hnmp) open
> # btrfs send -p /mnt/send/snapshots/2 /mnt/send/snapshots/3 | nc -v -c  127.0.0.1 6790
> At subvol /mnt/send/snapshots/3
> localhost [127.0.0.1] 6790 (hnmp) open
> # btrfs send -p /mnt/send/snapshots/3 /mnt/send/snapshots/4 | nc -v -c  127.0.0.1 6790
> At subvol /mnt/send/snapshots/4
> localhost [127.0.0.1] 6790 (hnmp) open
> write(net): Connection reset by peer
> 
> 
> 

Hi,

Haven't used netcat for btrfs send, so I am not sure why you have these 
issues. Have you tried mbuffer instead? This is what I use when I don't 
need to go over ssh.

Mbuffer[1] allows you to specify remote host and port. It also has an 
async buffer so it should be a bit faster too. In fact, it is useful on 
a localhost too, for example when sending snaps to a slower local backup 
media.

Receiving side:
# mbuffer -I port | btrfs receive /mnt/rec/snapshots

Sending side:
# btrfs send /mnt/send/snapshots/0 | mbuffer -O remotehost:port

Good Luck!

//Forza

[1]https://www.maier-komor.de/mbuffer.html