Search results

just add test repo,there is zfs 2.4

Even though I am in Europe not in EU, this is a great initiative. And I agree on Proxmox influence on virt services.

ZFS 2.4 is in test atm. And it is working good at >5 machines atm.

My recommendation is to use physical links to corosync, not vlan ones. As for the other things this is okay, all links should be active-passive in proxmox, so that you don't care if something dies ,and this is it. CEPH will work great in that case, those ssds work really great in practice.

Then okay, give them one disk(ceph) for os and one for VM and that is that. No raid inside and other bs.

All of this is okay, but how will you run ports passthrough?

Yes, vm-1-hypervisor. Strange but okay if he got money for it. Why not?

I mean it makes sense if you sell something like proxmox with 2-4cpus and 64gb of ram, so that you can spin 20 pms

Okay you offer customers an hypervisor. It makes sense, not much but it does. I would always use CEPH for that.

It is okay offering, i see there is financial reason for it. Unfortunately only nested proxmox is an option. But i cannot give you performance loss. Maybe install os on raid somewhere and add 2xnvme as passthrough for vmdata.

He is already using ssds for it's pool. You can add mirrors to mirrors ,effectively getting raid10.

Maybe in paranoid case you can go with raidz3, so that 3 disks can die? And set up spare .

For stable cluster CEPH is the only way to go.

I've worked with a few companies who migrated huge load of RDSses so my recommendation is start with 3-node CEPH, 10g networking and work from that. Go up to let's say 10-15 nodes, then create new cluster. No problem with that.

I circumvent that on supermicro with shutting down all vm and ct and then updating/upgrading everything. Try that.

Usually with NMS or monitoring systems in big support companies, you have one machine outiside of everything(different power,switch and 3g modem usually) so that when anything or everything dies you get notifications etc. If you are maintaining more than 10-20 clusters than it makes sense to...

No,didnt need that on my end.

here is how i do it in ex2200: ge-0/0/21 { description SP1-data; unit 0 { family ethernet-switching { interface-mode trunk; vlan { members [ Server-Vlan Host-Vlan Voice-Vlan Wifi-Vlan ];

I had similar problem with megaraid_sas, the zfs 1 boot disks couldnt write to it if the machine load was high. Once i've shutdown the Vms on it, the kernel upgrade or proxmox-boot-tool would run okay. this was supermicro.

Usually very high load shows problems with storage, eg disks cannot write fast enough. First look at disks.