No, I have not said his current hardware is trash but I think you have forgotten to analyze the hardware at hand! Each node has a single disk, presumably HDD so given any advice using this for HA will only lead to a frustrated user blaming proxmox for a bad solution. Running HA on a single disk on each node will be dead slow and only cause grief and disappointment. This is the reason I proposed buying a dedicated storage node.
sorry to say it mir, but that is what you sounded like
.
"do as I say or don't do it at all" >:-(
however with OP available hardware he can in fact run a somewhat HA setup.
not perfect by no means but doable.
if I was to fix your response, I would say to OP something like,
"
Hello <OP>, while the hardware you already have is capable of running HA configuration, it would not be an optimal
setup and the whole thing would be slow and difficult to setup and manage.
if you want to try and set it up anyway here what you do: <insert the possible setup outline here>.
However if you need/want a robust and workable configuration, I would add a shared storage like a dedicated qnap or similar unit to the mix, if budget allows it. "
This would have been a nice and friendly response that also provides a useful information to OP.
now that said, I would like to add my 2 centse to the thread.
@
Lirio > you do have almost enough hardware to setup and run a nice HA Proxmox cluster.
to outline the things :
A. HA Proxmox cluster require 3 nodes. while it is possible to run it on 2 nodes it will not be a smooth sailing.
B. any kind of HA setup needs an external shared storage. or at very least a decent specks on member nodes when it comes to network and storage.
i.e. if you would go with what you have with no external shared storage, you would need at least couple of extra NIC cards in each node to use solely for storage synchronization/replication use as in at least 2 nic cards to be used only for DRDB or Ceph to move/replicate data between the nodes real time fast enough for all to run smoothly.
you might also want to get a dedicated NIC card for cluster management useage. so to recap,
rach node might need at least 4 NICs if you go with 1GB speed.
1 for management
1 for network usage (this is the one you will use to connect all VMs to the network etc. while you can use the same port for management it is not recommended for HA setup. )
2 or more nics dedicated for shared storage use (Cifs or drdb)
if you can you can have 1-2 1GB ports for all network usage and get a 10Gb cards to use for storage nees.
now with dedicated 10Gb ports you can run Cifs/drdb setup on what you have quite comfortably. NOT perfect but comfortably, at least until your needs overtake the capabilities of the hardware.
if budget permits, in addition to what was said above, i would add an SSD to each node to install OS and run the VMs from, leaving the slower spinner totally for data storage.
and would plan to maybe adding second spinner and converting to at least RAID-1 setup, using ZFS or whatever to make the nodes more stable and data exchange faster.
if you want to make all this even more stable, plan for raid-1 OS install.
this way if OS disk fails you can just replace one and keep going. Proxmox VE 4.3+ support install on ZFS raid-1 directly.
