Truenas SCALE "system" snapshots (100-dreds after a few weeks)

[TueO815]

Hi,
I realized that after a few weeks running latest Truenas Scale 22.02. (RC2 and Nightlies) with 2 Apps (Jellyfin & Komga) a s well as one VM (Ubuntuserver) my Snapshot"folder" is filling up with hundreds of "cryptic" snapshots of aparently some dockerfiles and others - I guess from the installed apps. Like e.g.: MyPOOL/x-applications/docker/00907964e0ccc92212ff80adf73.................................

Trying to delete them does not work for most of them as they seem to depend of each other.

It mainly disturbes me for a few reasons:
a. to see them at all in case these are necessary "system" snapshots (would be nice to be able to hide/unhide in that case)
b. not beeing able to find easily my own snapshots because of the huge number
c. that I can not delete properly - as connections/dependencies are not shown.

Any solution for this ?

Thank's

 

[mkarwin]

I've also been experiencing this, only in my case, after the upgrade from previously somewhat well functioning BETA2 to RC2 to RELEASE, the performance hit possibly stemming from these is quite major - nearly all storage related actions take really long, several GUI/WebUI pages are loading for ages or are not presenting any data, boot seems to take not minutes but hours and days for pools import job.
I've discussed my case in https://www.truenas.com/community/threads/housekeeping-functionalities-in-scale-22-02-0-release-still-missing-thus-impacting-perceived-performance.99964/

 

[NugentS]

I suggest logging this as a bug in JIRA

 

[mkarwin]

I suggest logging this as a bug in JIRA

I don't think this is a bug per se, it's just the fact that there are so many snapshots created automatically by TrueNAS under ix-applications dataset that the performance drops a lot as the server is being actively used/updated, drops so much that a reboot/pool import/service startup/app deployment gets impacted very heavily and actions that should take seconds or minutes are taking minutes and hours/days instead. It might not be as discernible when running on pretty powerful/beefy hardware, but a lot of the people will be running the system on pretty limited hardware - more NAS-appliance-like than compute-node-server.
As such I've added this issue/request as suggestion in JIRA hoping someone will deal with it and the fix will get planned soon eg. for release 2022.02.1.