Replication Task Fails when remote "canmount=noauto" (but canmount=off works!)

[Richard Durso]

Running TrueNAS-12.0-U1.1

I'm using the Replication Task to replicate my encrypted periodic snapshots to a remote Ubuntu 20.04.1 system via SSH. This all works wonderful..... until I need to reboot the Ubuntu server. Upon reboot it will prompt me for the password of every replicated dataset (which I don't want). The backups are in the"rpool" pool which I think is part of the problem.

As a work around, I set all the child replicated datasets of "rpool/backups" to "canmount=noauto". Rebooted and password issue fixed. It only asks me for the ZFS on Root password and boots. Then to inspect the data when needed, I just use a simple "zfs load-key rpool/backups/<dataset>" and "zfs mount rpool/backups/<dataset>" and I can view the backups.

This seemed great until I started getting system alerts from TrueNAS that the replications failed with "Error 'noauto' " (see below). For now I set "canmount=off" for the the replicated datasets and all the replicated tasks cleared and seem to have no problem with "off". I assume this is a bug and it just doesn't expect "noauto".

Code:

Error
'noauto'.
Logs
[2021/01/27 12:00:00] INFO     [Thread-2739] [zettarepl.paramiko.replication_task__task_3] Connected (version 2.0, client OpenSSH_8.2p1)
[2021/01/27 12:00:00] INFO     [Thread-2739] [zettarepl.paramiko.replication_task__task_3] Authentication (publickey) successful!
[2021/01/27 12:00:02] INFO     [replication_task__task_3] [zettarepl.replication.run] For replication task 'task_3': doing push from 'main/docker' to 'rpool/backups/docker' of snapshot='auto-20210127.1200-1h' incremental_base='auto-20210127.1100-1h' receive_resume_token=None encryption=False
[2021/01/27 12:00:04] INFO     [replication_task__task_3] [zettarepl.replication.run] For replication task 'task_3': doing push from 'main/docker/projects' to 'rpool/backups/docker/projects' of snapshot='auto-20210127.1200-1h' incremental_base='auto-20210127.1100-1h' receive_resume_token=None encryption=False
[2021/01/27 12:00:05] INFO     [replication_task__task_3] [zettarepl.replication.run] For replication task 'task_3': doing push from 'main/docker/projects/gitea_data' to 'rpool/backups/docker/projects/gitea_data' of snapshot='auto-20210127.1200-1h' incremental_base='auto-20210127.1100-1h' receive_resume_token=None encryption=False
[2021/01/27 12:00:07] INFO     [replication_task__task_3] [zettarepl.replication.run] For replication task 'task_3': doing push from 'main/docker/projects/nzb_data' to 'rpool/backups/docker/projects/nzb_data' of snapshot='auto-20210127.1200-1h' incremental_base='auto-20210127.1100-1h' receive_resume_token=None encryption=False
[2021/01/27 12:00:08] INFO     [replication_task__task_3] [zettarepl.replication.run] For replication task 'task_3': doing push from 'main/docker/projects/trilium_data' to 'rpool/backups/docker/projects/trilium_data' of snapshot='auto-20210127.1200-1h' incremental_base='auto-20210127.1100-1h' receive_resume_token=None encryption=False
[2021/01/27 12:00:10] ERROR    [replication_task__task_3] [zettarepl.replication.run] For task 'task_3' unhandled replication error KeyError('noauto')
Traceback (most recent call last):
  File "/usr/local/lib/python3.8/site-packages/zettarepl/replication/run.py", line 154, in run_replication_tasks
... 6 more lines ...
    return [
  File "/usr/local/lib/python3.8/site-packages/zettarepl/dataset/list.py", line 33, in
    {
  File "/usr/local/lib/python3.8/site-packages/zettarepl/dataset/list.py", line 34, in
    property: parse_property(value, properties[property])
  File "/usr/local/lib/python3.8/site-packages/zettarepl/transport/zfscli/__init__.py", line 105, in parse_property
    return type(value)
  File "/usr/local/lib/python3.8/site-packages/zettarepl/transport/zfscli/parse.py", line 10, in zfs_bool
    return {
KeyError: 'noauto'

Now I have to do three steps instead of two to inspect the replicated copy:

Code:

zfs set canmount=on rpool/backups/<dataset>
zfs load-key rpool/backups/<dataset>
zfs mount rpool/backups/<dataset>

Then I can view the backups... and of course the 3 equivalent steps to undo the above so really 6 steps per dataset.