Unwanted hourly snapshots and related Syncoid errors

OpenZFS Sanoid
1

I use sanoid on “nas1” and syncoid to replicate, irregularly, to “nas2”. Typically this is once per week when I power on nas2. I’m asking for some help to understand why I have hourly snapshots/bookmarks on nas1, with what I believe to be an associated error during replication: “could not find any snapshots to destroy; check snapshot names”. This doesn’t affect all datasets, just those for which I do not have hourly snapshots configured,

From my sanoid.conf (these options have been unchanged for months):

[tank/archive]
        weekly = 4 
        monthly = 6
        yearly = 1
        autosnap = yes
        autoprune = yes

Then, my syncoid cli options:

/usr/sbin/syncoid --delete-target-snapshots --no-privilege-elevation \
        --no-sync-snap --create-bookmark --sshkey /home/name/.ssh/syncoid tank/archive name@nas2:tank/archive

Despite these settings I still have hourly snapshots for this dataset on nas1:

<snip>
tank/archive@autosnap_2024-05-25_00:00:09_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_01:00:02_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_02:00:03_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_03:00:05_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_04:00:02_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_05:00:02_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_06:00:01_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_07:00:06_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_08:00:03_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_09:00:00_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_10:00:03_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_11:00:03_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_12:00:02_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_13:00:02_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_14:00:04_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_15:00:03_hourly      0B      -  3.07G  -
tank/archive@autosnap_2024-05-25_16:00:05_hourly      0B      -  3.07G  -
<snip>

During replication to nas2, I receive these errors:

NEWEST SNAPSHOT: autosnap_2024-05-25_15:00:03_hourly                                                                                                                                         
Sending incremental tank/archive@autosnap_2024-05-18_00:00:08_daily ... autosnap_2024-05-25_15:00:03_hourly (~ 34 KB):
35.0KiB 0:00:00 [ 117KiB/s] [==============================================================================================================================================] 102%            

ssh     -i /home/name/.ssh/syncoid -S /tmp/syncoid-name@nas2-1716650406-3134 name@nas2 ' zfs destroy '"'"'tank/archive'"'"'@autosnap_2024-05-18_10:00:04_hourly,autosnap_2024-05-18_11:00:07_
hourly,autosnap_2024-05-18_12:00:00_hourly,autosnap_2024-05-18_13:00:02_hourly,autosnap_2024-05-18_14:00:03_hourly,autosnap_2024-05-18_15:00:01_hourly' failed: could not find any snapshots to destroy; check snapshot names.

On nas2, the excerpt of snapshots for the same period of time looks like this:

tank/archive@autosnap_2024-05-25_00:00:09_daily       0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_00:00:09_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_01:00:02_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_02:00:03_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_03:00:05_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_04:00:02_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_05:00:02_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_06:00:01_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_07:00:06_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_08:00:03_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_09:00:00_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_10:00:03_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_11:00:03_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_12:00:02_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_13:00:02_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_14:00:04_hourly      0B      -  3.08G  -
tank/archive@autosnap_2024-05-25_15:00:03_hourly      0B      -  3.08G  -

As I say, with no hourlies configured for sanoid, I am failing to understand how I am setting the conditions for this issue and the error to occur. Thank you in advance for any time you’re willing to spend helping me with this :slight_smile:

2

What version number is your syncoid? Delete-target-snapshots is a feature I actually fought against for a few years before it finally snuck in; I’m not sure when it arrived but it’s possible it might just not be there in the version you’ve got?

3 
└──╼ $syncoid --version
/usr/sbin/syncoid version 2.2.0
(Getopt::Long::GetOptions version 2.52; Perl version 5.36.0)

The error is something but it might not be the root issue, which was why are there hourly snapshots in the first place, given my sanoid.conf settings.

4

For future searchers, I think I’ve solved it. Since I wasn’t using a template and hadn’t specified the number of snapshots of each type (I only specified weekly, monthly, yearly) it seems that sanoid was pulling a default value of “hourly = 48” from somewhere and applying it. It was doing something similar for daily as well. Having amended my sanoid.conf to look like this:

[tank/archive]
	hourly = 0
	daily = 0
	weekly = 4 
	monthly = 6
	yearly = 1
	autosnap = yes
	autoprune = yes

It seems to function as I had hoped. It was this Note in the wiki which helped:

Note: Be aware that if you don’t specify some interval options the defaults will be used (from /etc/sanoid/sanoid.defaults.conf)

However, I’m not sure where the apparent default of 48 is listed in the sanoid.defaults.conf file.

1 Like
5

The configs fall down as follows, in order of overriding:

  • sanoid.conf (from last to first)
  • defaults.sanoid.conf
  • hardcoded defaults in sanoid itself

So if a value is specified in sanoid.defaults.conf (which should be the case for all global configs), it overrides the hardcoded defaults in sanoid itself.

If another conflicting value is specified in sanoid.conf, it overrides the one in sanoid.defaults.conf.

If another value further down the path in sanoid.conf–eg, daily=38 in the [parent/child] module, and daily=37 in the [parent] module–the one further down overrides the one further up, so in this instance the value for daily in [parent/child] would be 38, not 37, since (regardless of what order you place them in sanoid.conf itself) [parent/child] is further down the hierarchy than [parent].

3 Likes