Zfs send & recieve eats my memory and kills vms

I used zfs send / recieve to transfer a whole dataset but obviosly this eats all my memory so that finally all VMs are dead. What can i do ?

Trusnas Core Version: TrueNAS-13.0-U6.1

Can someone please help to understand this behavior.
is homedir something special?
Why truenas is not able to transfer this big bunch of data from one to the other pool without running into swap and filling the memory for something… ?
I killed the zfs send + receive but receive still running with 100% on one core.

last pid: 35599; load averages: 5.24, 4.32, 2.64 up 0+02:03:54 15:56:43
85 processes: 2 running, 83 sleeping
CPU: 0.1% user, 0.0% nice, 9.0% system, 0.0% interrupt, 90.9% idle
Mem: 13G Active, 4096B Inact, 11G Laundry, 5483M Wired, 545M Free
ARC: 1002M Total, 528M MFU, 52M MRU, 268M Anon, 102M Header, 50M Other
792M Compressed, 890M Uncompressed, 1.12:1 Ratio
Swap: 6144M Total, 6144M Used, K Free, 100% Inuse

swap is filled after ~ 17G transfered and the whole system runs into trouble afer the swp is filled for what ever reason

The swap not working properly is an issue, but I think that, fundamentally, you do not have enough RAM. 16 GB is the recommend minimum for TrueNAS using storage alone; further workload, including VMs, requires more RAM.

I do have 32Gig of RAM and the VMs configured to use 4GB each

Same issue with all VMs stopped before starting the task

The official minimum is 8GiB