[SOLVED] AMD EPYC 7502P Support
- Thread starter djsami
- Start date
hi,
if i get it right, you're asking how many centos VMs you can run on this node?
that all depends on the resources you assign to each VM.
note that PVE needs some amount of resources to function by itself (you can't assign everything to VMs). this need gets bigger if you're going to use something like ZFS which needs a lot more RAM.
if i get it right, you're asking how many centos VMs you can run on this node?
that all depends on the resources you assign to each VM.
note that PVE needs some amount of resources to function by itself (you can't assign everything to VMs). this need gets bigger if you're going to use something like ZFS which needs a lot more RAM.
hi,
okay, so you know how each VM will be allocated, you need to calculate simply:
32 cores / 4 cores = 8 VMs
256G ram / 4G ram = 64VMs but this is capped at 8 because of the cores. each VM needs at least 1 core minimum so you get 32VMs max.
which means based on the cores you have between 8-32 VMs you can run on the node. space is also no problem with almost 5TB if you're gonna give 100G for each VM (5000 / 100 = 50 but that's over 32 already).
hope this helps!
okay, so you know how each VM will be allocated, you need to calculate simply:
32 cores / 4 cores = 8 VMs
256G ram / 4G ram = 64VMs but this is capped at 8 because of the cores. each VM needs at least 1 core minimum so you get 32VMs max.
which means based on the cores you have between 8-32 VMs you can run on the node. space is also no problem with almost 5TB if you're gonna give 100G for each VM (5000 / 100 = 50 but that's over 32 already).
hope this helps!
i want to give further info in addition to my colleague
so here the limit mostly has to do with what your vms actually do
available on the real machine
(especially if your vms are all the same)
you to give 'more' disk space to vms than you actually have
caveats apply though:
in any situation where you overcommit: monitor your hosts
especially for disks space this is essential, because when you overcommit you 'promise' that not all vms
use all the resources at the same time, but when they do, everything either crashes, come to a halt or in the worst
case, you can lose data!
its 64 threads, and in reality most assigned cpu cores of vms are not doing much, cpus are very easy to overcommit
so here the limit mostly has to do with what your vms actually do
again not so easy, if you enable ballooning and/or ksm you might be able to overcommit to much more than you have
available on the real machine
(especially if your vms are all the same)
also this is (depending on the storage) not necessarily true, since you can use a 'thin-provisioned' storage which allows
you to give 'more' disk space to vms than you actually have
caveats apply though:
in any situation where you overcommit: monitor your hosts
especially for disks space this is essential, because when you overcommit you 'promise' that not all vms
use all the resources at the same time, but when they do, everything either crashes, come to a halt or in the worst
case, you can lose data!
anlamıyorum oguz.I asked 5 times? oguz You did not answer 5 times what is your problem?
oguz name is a Turkish name
What is your problem? This forum language is English, we do not have a Turkish forum, I'm afraid.
And just because a name sounds like it comes from some country it doesn't mean that they necessarily have to be from that country..
I'm so sorry, you just asked 1 2 times I wondered why he did not answer, we are a nationalist, I told him in that respect