Veeam Backups - VMs are getting STUNNED on snapshot removal

  • 23 November 2016
  • 3 replies

Hi All
We are just bringing a new Nutanix Block online and starting to move some production VMs.With our Veeam backup testing we are seeing noticeable stun (in some cases up to a minute of VM unavailable) issues with snapshot removals that are done as part of Veeam backups. We have a veeam proxy installed on the nutanix NFS datastore, and are running vsphere6, latest Nutanix OS, etc. and latest version of Veeam 9.5.I did find a article on Nutanix forums about forcing NDB/network mode to help with this and this did improve thing but not enough to call it resolved. I cannot have my primary SQL server stunned every time we do a backup, so this is a bit of show stopper at this point.Just looking for ideas at this point?Jeff

This topic has been closed for comments

3 replies

Userlevel 4
Badge +19
I think this blog post might help you

NFS with the apporaite regkeys
Userlevel 6
Badge +29
dlink7's article is great, run through that and see how it goes.

If it doesn't help, please open a support ticket with both us and Veeam, we're happy to dig into the details collaboratively.
Userlevel 6
Badge +29
Also, its worth saying that remote hot add and the long stun times is a VMware issue that's been in VMware's product for literally the better part of a decade:

They also recommend LAN/NBD mode for VADP enabled backup products, like Veeam. I've seen this exact same issue with Veeam, NetWorker, Avamar, NetBackup, and so on, and LAN/NBD mode almost always solves it.

The Veeam reg keys that dlink7 mentioned gives you some opportunistic HotAdd, which is historically "faster" than NBD mode (with any storage, ever, just because of the way the IO for NBD mode), if the Veeam proxy and the target VM are on the same host at the same time. a nice to have, but not perfect.

I haven't personally used DirectNFS from Veeam yet, which is a feature in v9+, so no idea if that will help here or not.