3 ms·

I'm already not going to renew my VMUG membership, and am planning to switch my VSphere cluster over to Proxmox.

by wpm 3y ago

helf 3y ago
[dead]
justinclift 3y ago
Are you using live migration of running VMs, or wanting it to work in Proxmox? Asking because in my testing of that in Proxmox 3 and 5 node clusters last month, the live migration would fail (aka just hang with no further progress) just over 20% of the time. Haven't looked into it further (likely will though), but searching around online shows lots of other people having experienced the same issue. General sentiment seems to be "Proxmox is good for homelab and non-prod environments. But probably don't use it in production."
- lightbritefight 3y ago
  Check put XCP-NG if youre looking in this space. Its a mature xenserver fork with a much improved interface. Migrations work well, and it has a comparatively inexpensive vsan equivalent that works well if you prefer a hyperconverged route.
  - justinclift 3y ago
    Thanks. I looked at it briefly, but the project doesn't seem to be headed in a good direction. The main interface (Xenadmin) now has an EOL notice as their README: https://github.com/xcp-ng/xenadmin https://github.com/xcp-ng/xenadmin The other alternative now seems to be a paid solution: https://vates.tech/blog/introducing-vates-virtualization-management-stack/ https://vates.tech/blog/introducing-vates-virtualization-man... That doesn't seem like a good match for my needs. :(
    - tecleandor 3y ago
      I think you can use Xen Orchestra with xcp-ng, isn't it?
      - justinclift 3y ago
        That's the second link I think. At least, when going to their web page, that 2nd link is what they point to.
- edward28 3y ago
  Perhaps the performance of your storage or network isn't up to scratch. I have a 3 node cluster with ceph on ssds and a 10gb network and my live migrations worked perfectly fine.
  - justinclift 3y ago
    This was all in a virtualised setup on my local workstation (Ryzen 5950X, 64GB ECC, local mirrored NVMe SSDs). I've not seen flaky comms between virtual interfaces on this setup before, but you never know. ;)

I'm already not going to renew my VMUG membership, and am planning to switch my VSphere cluster over to Proxmox.

by wpm 3y ago

I'm already not going to renew my VMUG membership, and am planning to switch my VSphere cluster over to Proxmox.

helf 3y ago

[dead]

justinclift 3y ago

Are you using live migration of running VMs, or wanting it to work in Proxmox? Asking because in my testing of that in Proxmox 3 and 5 node clusters last month, the live migration would fail (aka just hang with no further progress) just over 20% of the time. Haven't looked into it further (likely will though), but searching around online shows lots of other people having experienced the same issue. General sentiment seems to be "Proxmox is good for homelab and non-prod environments. But probably don't use it in production."

lightbritefight 3y ago
Check put XCP-NG if youre looking in this space. Its a mature xenserver fork with a much improved interface. Migrations work well, and it has a comparatively inexpensive vsan equivalent that works well if you prefer a hyperconverged route.
- justinclift 3y ago
  Thanks. I looked at it briefly, but the project doesn't seem to be headed in a good direction. The main interface (Xenadmin) now has an EOL notice as their README: https://github.com/xcp-ng/xenadmin https://github.com/xcp-ng/xenadmin The other alternative now seems to be a paid solution: https://vates.tech/blog/introducing-vates-virtualization-management-stack/ https://vates.tech/blog/introducing-vates-virtualization-man... That doesn't seem like a good match for my needs. :(
  - tecleandor 3y ago
    I think you can use Xen Orchestra with xcp-ng, isn't it?
    - justinclift 3y ago
      That's the second link I think. At least, when going to their web page, that 2nd link is what they point to.
edward28 3y ago
Perhaps the performance of your storage or network isn't up to scratch. I have a 3 node cluster with ceph on ssds and a 10gb network and my live migrations worked perfectly fine.
- justinclift 3y ago
  This was all in a virtualised setup on my local workstation (Ryzen 5950X, 64GB ECC, local mirrored NVMe SSDs). I've not seen flaky comms between virtual interfaces on this setup before, but you never know. ;)