VM scheduling problem

Hello, I have by default configured

DEFAULT_SCHED = [
    policy = 1
]

which is policy “striping”, but scheduler behaves like policy “packing” (0) is configured. New VMs are deployed to host with most allocated CPU and MEMORY.

There is no scheduling defined in VM template (except DS policy, which works fine).

What can cause this behaviour? Thanks


Versions of the related components and OS (frontend, hypervisors, VMs):
5.4.11

Can you paste here your sched.conf and your VM Template ?

I’m trying to reproduce this problem on my ONE but I can’t.

regards

Hello, thank you for reply. Meanwhile I changed cluster configuration and updated to 5.6. Before I have heterogenous cluster - each node have different CPU family and core count. I have homogenous cluster now. So probably was caused by that or it is fixed in 5.6.

Ok, thanks for feedback!