Clarification needed: NFS, Snapshots and stunning

VMware specific discussions

Clarification needed: NFS, Snapshots and stunning

Veeam Logoby svenh » Thu Feb 04, 2016 9:35 am

Hi!

With Veeam v9 including the new DirectNFS Client, I wonder about the following situation:

Prior to v9, Veeam could only do HotAdd or NBD if your datastores were on NFSv3. Because of a locking problem with NFSv3, a VM could get stunned for a long amount of time, if the HotAdd proxy was not on the same host as the VM (https://www.veeam.com/kb1681 and http://kb.vmware.com/kb/2010953).

Solutions were to either only use NBD (which can be slow and imposes a greater burden on the ESX) or to deploy a proxy to each host and set EnableSameHostHotaddMode=2 in the registry on the Veeam Backup server.

Now my question: Has this problem dealt with with the new DirectNFS client?
Second question: Is a Veeam Enterprise license sufficient, if I don't need Storage Snapshot integration to be able to use DirectNFS?

Thanks in advance,
Sven.
svenh
Enthusiast
 
Posts: 26
Liked: 5 times
Joined: Thu Jan 21, 2016 12:56 pm
Location: Germany
Full Name: Sven Hartge

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby PTide » Thu Feb 04, 2016 11:23 am 1 person likes this post

Hi,

Now my question: Has this problem dealt with with the new DirectNFS client?
Depending on NFS server version v9 NFS client uses either v3 or 4.1. Since in both cases (Hot-Add with proxy on another host and DirectNFS with proxy on another machine) your VM disks are mounted to some other machine the locking issue persists for NFS v3. If your datastore allows NFS v4 then no locking issue should occur.

05.02.2016 UPDATE: Please disregard what's been written about locks. I checked with devs and they clarified that Veeam Direct-NFS transport mode will not result in VM stuns. Although the scenario might look somewhat similar to off-host hot-add proxy direct-NFS works differently and does not cause lock issues.

Is a Veeam Enterprise license sufficient
Yes, Enterprise is sufficient. Moreover, there are no Transport mode limitations depending on license, all transport modes are available in all editions.

Thank you.
PTide
Veeam Software
 
Posts: 3019
Liked: 246 times
Joined: Tue May 19, 2015 1:46 pm

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby svenh » Thu Feb 04, 2016 11:34 am

PTide wrote:
Now my question: Has this problem dealt with with the new DirectNFS client?

Depending on NFS server version v9 NFS client uses either v3 or 4.1. Since in both cases (Hot-Add with proxy on another host and DirectNFS with proxy on another machine) your VM disks are mounted to some other machine the locking issue persists for NFS v3. If your datastore allows NFS v4 then no locking issue should occur.

Is this interpretation correct: Since I am using ESX 5.5 with NFS on NetApp Ontap 7-Mode I am limited to NFSv3, which means DirectNFS does nothing for me and I have to deploy a proxy on every ESX and use HotAdd (or use NBD) to avoid stunning. Only after I upgrade to ESX6 I can use NFS4 and utilize DirectNFS without HotAdd.
svenh
Enthusiast
 
Posts: 26
Liked: 5 times
Joined: Thu Jan 21, 2016 12:56 pm
Location: Germany
Full Name: Sven Hartge

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby PTide » Thu Feb 04, 2016 12:02 pm 1 person likes this post

Yes, everything is correct.

05.02.2016 UPDATE: Please disregard that and see this post.
PTide
Veeam Software
 
Posts: 3019
Liked: 246 times
Joined: Tue May 19, 2015 1:46 pm

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby svenh » Thu Feb 04, 2016 12:05 pm

PTide wrote:Yes, everything is correct.


This is unfortunate, I thought DirectNFS solves this problem for me. Well then, time to prepare some proxy servers.
svenh
Enthusiast
 
Posts: 26
Liked: 5 times
Joined: Thu Jan 21, 2016 12:56 pm
Location: Germany
Full Name: Sven Hartge

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby PTide » Thu Feb 04, 2016 12:14 pm

Though the issue is known to occur in some environments I still suggest you to perform some tests in your environment. There are other factors besides NFS that trigger the problem (I/O load) and it might be so that everything will work fine for you.

Thank you.
PTide
Veeam Software
 
Posts: 3019
Liked: 246 times
Joined: Tue May 19, 2015 1:46 pm

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby svenh » Thu Feb 04, 2016 12:28 pm

PTide wrote:Though the issue is known to occur in some environments I still suggest you to perform some tests in your environment. There are other factors besides NFS that trigger the problem (I/O load) and it might be so that everything will work fine for you.


I will of course experiment with this in a pre-production environment, but in the end I cannot risk any long-time stunning of VMs and will err to the safe option of either doing NBD or deploying a proxy on every ESX host.
svenh
Enthusiast
 
Posts: 26
Liked: 5 times
Joined: Thu Jan 21, 2016 12:56 pm
Location: Germany
Full Name: Sven Hartge

Re: Clarification needed: NFS, Snapshots and stunning

Veeam Logoby svenh » Fri Feb 05, 2016 2:50 pm 2 people like this post

PTide wrote:05.02.2016 UPDATE: Please disregard what's been written about locks. I checked with devs and they clarified that Veeam Direct-NFS transport mode will not result in VM stuns. Although the scenario might look somewhat similar to off-host hot-add proxy direct-NFS works differently and does not cause lock issues.


Very nice, thank you for checking.

I think it could be worthwhile to add this information to KB1681.
svenh
Enthusiast
 
Posts: 26
Liked: 5 times
Joined: Thu Jan 21, 2016 12:56 pm
Location: Germany
Full Name: Sven Hartge


Return to VMware vSphere



Who is online

Users browsing this forum: Exabot [Bot] and 27 guests