vSphere 5.5 tests done?

[Gostev]

It canceled and failed over to network mode. In exactly this second the whole ESX gives me a purple screen of death! I thought it was just an error. Tried again, and it happened again just in this second where it failed over to network mode. So something is really not working there.

I concur. We've been doing stress testing of our vSphere 5.5 support code for the past few weeks, and we've seen plenty of PSODs with ESXi 5.5 under heavy load... it is prone to crashing like no other ESXi version before. I like to say, Windows Vista was solid as a rock comparing to ESXi 5.5 (and this says a lot).

Of course, we are using all the same hosts that were previously running older ESXi versions, and those were solid. So, things are not looking good for ESXi 5.5.

We have some ideas on what specifically is causing this, confirming them right now by eliminating a suspect from the equation, and will be opening a support case with VMware soon.

P.S. Specifically regarding hot add not working, this is the known interoperability issue between vSphere 5.5 and VDDK 5.1 that is resolved in VDDK 5.5.

[dkvello]

Hmm, I've got some 120 hosts running on 5.5 (on mulitple sites, IBM and HP Hardware) and have not seen any PSOD's or other instability. All sites run Fiber-channel storage (IBM Storwize, EmC VNX, HP 3PAR).
I've only encountered API problems with Veeam 7 against ESXi 5.5 hosts when using SAN based backup, so I've moved to Network for the time being.
What exacttly do You suspect being the culprit in your PSOD endeavours?

[Gostev]

E1000E vNIC under heavy load.

[Vasko]

E1000E vNIC under heavy load.

I maybe have a confirmation on weird E1000E behavior:

I'm building a new AD Environment based on ESXi 5.5. Default Adapter when adding NIC to Machine in vSphere WebClient 5.5 is E1000E which I didn't change.
In the process of investigating sync issues between two DCs/Exchange Servers, I figured out that every packet that's been received/send by an E1000E has its last 8bytes set to ZEROS. Disabling FlowControl and/or Checksum offload didn't help a bit.
After I changed the Adapter to VMXNET3 the Problems went away.

Maybe this helps your tests.

Greetings,
Vasko

[dkvello]

E1000E vNIC under heavy load.

Ah, I haven't used Flexible or E1000 for many years. Only VMXNET variants for as long as I can remember. No wonder I haven't run into any of those PSOD's or W2K12 + E1000 problems.

[Gostev]

The problem here is that E1000E is the default choice for the new VMs since vSphere 5.0.

[jsurcombe]

The E1000E NIC seems to cause a PSODs with Windows 2012 R2 or Windows 8.1 guests on (at least) vSphere 5.1 U1 and vSphere 5.5 when under network load. Changing the NIC to VMXNET3 is the recommended workaround until the next update of vSphere 5.5. I suspect a lot of people are going to hit this, because the E1000E NIC is the default for Windows 2012 guests (and there's no separate option for 2012 R2).

See https://communities.vmware.com/message/2297546

[jamerson]

i have installed Veeam 7 on ESXI 5.5 i've created a backup job,
it works really ( 2008R2 ), i backedup both DC and Exchange

[donativo]

If you can reproduce that condition on separate hardware, it seems you stumbled upon a nasty ESXi bug -- Even if Veeam is utterly broken, it should not be capable of crashing ESXi with API calls.

I do think the vSphere community could potentially benefit if you can get more info on the purple screen condition and report the purple screen issue to VMware, as you may have exposed a bug that also has security ramifications for ESXi.

i think vmware and its community figured this out already.
https://communities.vmware.com/message/2297546

I will also give it a try to change the NIC on my Server 2012 r2 VM to VMXNET3 or E1000. But unless Server 2012R2 is not supported anyway by Veeam I will wait.

[smuzzey]

i have installed Veeam 7 on ESXI 5.5 i've created a backup job,
it works really ( 2008R2 ), i backedup both DC and Exchange

I did the same and have been running it for a while. Hottadd doesn't work but everything else was working fine so on my small cluster so I upgraded to 5.5. For a couple of weeks it was running fine until Friday when it started giving me trouble. All but one of my VM's give me failed to open VDDK disk errors. The strange part is nothing was changed and one VM still backs up fine.

Anyway I put in a ticket and just as I thought they won't deal with it because it's 5.5 so I guess I'll be using VSphere data protection until Veeam gets a working version for us. The did mention late next month but that's subject to change.

[jp34]

Hi all,
I don't understand if the issue comes from vCenter or ESXi version...
Would veeam 7 work fine backing up VMs running on an ESXi 5.0 managed by a vCenter Appliance 5.5 ?
Thanks in advance.

[foggy]

Some customers have reported above the successful Veeam B&R operation on this kind of setup, however full official vSphere 5.5 support will be available in the upcoming Veeam B&R v7 R2 release only.

[Morat]

I did see somewhere that there would be a patch2 for Veeam7.0 and it would support VMWare 5.5 - any news from anywhere?
Cheers,
Morat

edit: clearly I missed a page

Does anyone know if R2 is expected any time soon? If it's going to be next year then I'll need to upgrade to VMWare 5.1 then 5.5 when I'd rather go from 5.0 direct to 5.5.

[foggy]

We decided to refer to the patch 2 as Veeam B&R v7 R2 release due to the amount of new functionality it brings. So Patch #2 and v7 R2 update is the same release that will be out in several days.

[Morat]

Hurrah \o/

thanks foggy.
Morat

[smuzzey]

Excellent this is good to hear

[jja]

Ah, can't wait to get this version.
We upgraded to v7 last friday, guess we should have waited 2 weeks more..
Ah well, all the backups seems to work for now.

Will the "R2" release be a full upgrade job?

-j

[foggy]

It will not be a new full product build, but an update that is applied to the current installation.

[scott.anderson]

When when when ?! Today or tomorrow?

[superdekster]

November as Gostev said

[antivir]

tests done:
http://www.veeam.com/kb_articles.html/kb1831

[jja]

I upgraded from v7 build x.690 to the R2 patch now.
Looks very good.
Now the web client plugin works as it should aswell

Good job!

-j

[Gostev]

Yep, vSphere 5.5 tests are now done so I am locking down this thread. Thanks everyone for your patience!