I am developing software to monitor zfs snapshots on host and send/receive to host. On locally created snapshots, the latest snapshot might normally be taken every five minutes. On the receive end of hourly send/receive replication between nodes the latest snapshot could be an hour old, then suddenly look like we have five minute snapshots backdated through the past hour.
I have naming conventions for PRIMARY/COPY volumes on my system to apply the proper age thresholds, but I’m trying to make the check tool generic. I can read pyznap setup files, but there’s probably a half dozen different snapshot/replicate tools in use.
I’m hoping there might be something in the zfs properties of the volume or snapshot related to the origin, but I’m not seeing it. I could have the user manually add some property to the backup volumes, but I’m trying to automate as much as possible. Any suggestions are welcome.
I can’t think of anything off the top of my head. Pick any given snapshot and zfs get all pool/set@snap, then scroll slowly through the output and see if you can find a property that does what you want–that’s all the properties, so either it’s there or it isn’t. 
1 Like
I did pool and snapshot before posting. Nothing I recognize tells me what I want. I was hoping there was more magic behind some of these fields than a muggle could guess from reading the names. I suspect I’m off to write some heuristics.
root@nas1 [~] Red Hat Enterprise Linux 9.5 (Plow) 09:59:31 Sun Feb 09
# zfs get all ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly
NAME PROPERTY VALUE SOURCE
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly type snapshot -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly creation Sun Feb 9 7:00 2025 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly used 0B -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly referenced 255G -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly compressratio 1.11x -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly devices on default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly exec on default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly setuid on default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly createtxg 62298053 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly xattr on default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly version 5 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly utf8only off -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly normalization none -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly casesensitivity sensitive -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly nbmand off default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly guid 13560627357842945247 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly primarycache all default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly secondarycache all default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly defer_destroy off -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly userrefs 0 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly objsetid 38514 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly mlslabel none default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly refcompressratio 1.11x -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly written 0 -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly logicalreferenced 282G -
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly acltype off default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly context none default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly fscontext none default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly defcontext none default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly rootcontext none default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly encryption off default
ZRaid/COPIES/NAS0/ISOS@pyznap_2025-02-09_07:00:06_hourly prefetch all default
Try “zfs get -s received”. This will only be useful if the send stream had properties, such as when using “zfs send -R”. Also, it is only useful if you have actually set properties on the send side rather than just getting inherited properties.
box1# zfs create rpool/fs
box1# zfs set me:fromhost=box1 rpool/fs
box1# zfs snapshot rpool/fs@1
box1# zfs send --raw -R rpool/fs@1 | ssh box2 zfs recv backup/fs
box2# zfs get -s received all backup/fs
NAME PROPERTY VALUE SOURCE
backup/fs me:fromhost box1 received
1 Like
I hadn’t thought of that. I don’t see any property now with a received status on the first COPY volume or snapshot I checked. I can add a property at one or both ends per your example, but at that point I should probably just check for the value of the property itself.
zfs set santa:copyflag=on ZRaid/COPIES
# zfs get santa:copyflag ZRaid/COPIES -r -t filesystem
NAME PROPERTY VALUE SOURCE
ZRaid/COPIES santa:copyflag on local
ZRaid/COPIES/KYLO santa:copyflag on inherited from ZRaid/COPIES
ZRaid/COPIES/KYLO/LIBVIRT_IMAGES santa:copyflag on inherited from ZRaid/COPIES
ZRaid/COPIES/KYLO/LIBVIRT_QEMU santa:copyflag on inherited from ZRaid/COPIES
ZRaid/COPIES/NAS0 santa:copyflag on inherited from ZRaid/COPIES
ZRaid/COPIES/NAS0/ISOS santa:copyflag on inherited from ZRaid/COPIES
...
1 Like
The flag above is redundant to the naming convention I use now, but a user is more likely to add a flag to use my tool than rework their naming conventions.