ZFS Replication LXC Container fails

[cpzengel]

Hi Guys,

since Version 4.x we have Replication Issues with Open-ZFS
For some Reason LXC on Promox uses acltype=posixacl

This Attribute can´t be found on FreeNAS 9.10

Some Replications work with Core Dump Error, others only send first Snapshot and fails on the other!

Do I need this Attribute?

Example working with Error

rpool/data/subvol-107-disk-1:
newest local snapshot: zfs-auto-snap_daily-2016-10-18-1925
newest remote snapshot: zfs-auto-snap_daily-2016-10-17-1925
receiving incremental stream of rpool/data/subvol-107-disk-1@zfs-auto-snap_daily-2016-10-18-1925 into cosalux_repl/subvol-107-disk-1@zfs-auto-snap_daily-2016-10-18-1925
internal error: Invalid argument
Abort (core dumped)
Error sending snapshot.
​

Example failing

receiving full stream of rpool/vms/pwmanager/subvol-142-disk-1@zfs-auto-snap_weekly-2016-09-11-0448 into externgo/vms/pwmanager/subvol-142-disk-1@zfs-auto-snap_weekly-2016-09-11-0448

internal error: Invalid argument

Abort (core dumped)

warning: cannot send 'rpool/vms/pwmanager/subvol-142-disk-1@zfs-auto-snap_weekly-2016-09-18-0448': Broken pipe

warning: cannot send 'rpool/vms/pwmanager/subvol-142-disk-1@zfs-auto-snap_daily-2016-09-19-1730': Broken pipe​

...... and so on

Any Ideas?

Cheers Chriz

 

[fabian]

set acl=0 on all the ZFS-backed volumes of all your containers you have on your zpool, restart the containers, then check with "zfs list -o name,acltype" that acls are off. now you should be able to replicate it to non-linux ZFS hosts.

you need it if you want ACLs to work

 

[cpzengel]

i set aclmode to off, still same behavior

rpool off
rpool/Backup off
rpool/ROOT off
rpool/ROOT/pve-1 off
rpool/Replica off
rpool/Replica/CONTAINER off
rpool/data off
rpool/data/subvol-102-disk-1 off
rpool/data/subvol-103-disk-1 off
rpool/data/subvol-104-disk-1 off
rpool/data/subvol-105-disk-1 off
rpool/data/subvol-107-disk-1 off
rpool/data/vm-100-disk-1 -
rpool/data/vm-124-disk-1 -
rpool/swap -
rpool/vms off
rpool/vms/ISPCONFIG off
rpool/vms/PYDIO off
rpool/vms/WIN off
root@pve240:~# cd /etc/cron.daily/
root@pve240:/etc/cron.daily# ./02zfs-backup

rpool/data/subvol-102-disk-1:
newest local snapshot: zfs-auto-snap_daily-2018-03-28-1925
cannot open 'cosalux_repl/subvol-102-disk-1': dataset does not exist
Sending initial snapshot.
receiving full stream of rpool/data/subvol-102-disk-1@zfs-auto-snap_weekly-2018-02-04-0548 into cosalux_repl/subvol-102-disk-1@zfs-auto-snap_weekly-2018-02-04-0548
internal error: Invalid argument
Abort (core dumped)
Error sending initial snapshot.

FreeNAS 11.1 U3 - PVE 5..1-46

 

[fabian]

you'd need to find out which part of the 'zfs send' stream it is choking on - I don't know whether that is possible on FreeNAS/FreeBSD without modifying ZFS. could you post the full send and receive command lines and the pool and dataset properties?