Recent content by barius

Finally I bought 4 x 4TB spinning drives and 2 x 900GB SSD for the Ceph on each node. I didn't want all the OSD on one node to die if just one SSD failed. So the DB+WAL of two OSDs share one SSD and the DB+WAL of the two remaining OSDs share the other SSD. I can't really say how it performs...

In theory it should be in some cases. It depends on your OS installer... It's not enough to just create a serial port on the VM. That is like installing on a physical PC that just happens to have a serial port - it will not automatically be used instead of the screen and keyboard. You have...

Hi, What you describe is called "memory overcommitting" and is certainly something that can be done for the reason you describe (not all VM active at the same time). However, it does have some risk: Mainly, the memory can only be used by one process at any time (the VMs are also processes on...

Hello, I am preparing to order hardware for a 3-node hyper-converged cluster. I won't bore you with exact configuration details but I would be thankful for some general advice on this point. To optimise performance with a limited budget (all SSD storage is not an option) i have read that it...