Wan Accelerator eating all hyper-v host memory

[frankive]

Hi.
Is it possible to set a limit of huch resource the wan accelerator can use of a hyper-v host? We have a few customers with HP Microserver (just a few users and not extreme performance..!) with 16 GB Ram. We give their only VM 8 GB Ram and the host itself 8 GB so Veeam also can be happy.
It seems like it is consuming all the Ram when it is working which also make the hyper-v host not responding very well. The VM is not affected.

I it possible to tell the Wan accelerator to f.ex max consume 4 GB while it is working?

[dellock6]

Hi Frank,
it means you installed the WAN accelerator role on the host, and not in the VM? AFAIK there is no such configuration, and usually the RAM assigned to the server running the WAN Accelerator is completely used. The best way I think is to have it running on a dedicated VM, softwares like this can indeed affect all the VMs if they are running at the host layer.

[frankive]

Can the wan accelerator be installed to a linux vm? these small customers usuallt running 2012 server essentials which mean they only have 1 vm in the license. And if I install it in the production vm, it will take all the ram alså so I guess that would not benefit any in this situation?

[dellock6]

No, WAN accelerator only works on Windows.
I got your point now, the only windows VM in this small branch office design is the hyper-v host itself, and the production VM.
I'm not aware if we have a registry key to limit the consumed ram, there are ways to set the "maximum working set" of a process in windows, but they are obviously totally unsupported by us, and I'm not even sure they can create worse issues to the WAN accelerator itself. for this reason I'm not giving you any link, you can google it around

Luca.

[frankive]

Thanks!

Maybe this could be a feature request..? Lots of small customers out there you know...

[dellock6]

Which one? LInux installation or limit on RAM consumption? The first is unlikely as almost all our components are designed to run on windows system, while I see a point in limiting the ram consumption in scenarios where other services are running together with the wan accelerator.

[Gostev]

It seems like it is consuming all the Ram

Can you please tell me what is the exact name of the memory metric that makes you think so? I assume you are looking at the Task Manager - what is the name of the column?

Thing is, Windows memory management is quite complex - and as far I as remember, prior to Windows 2012 the default memory metric in Task Manager was the most useless one. You want to use Memory (private working set) column instead to justify on real physical memory consumption by any process.

[frankive]

I will certainly come back with the information you are lloking for the next time the job is running.
But yes, I saw just at the task maanger and saw that the wan accelerator was using 7GB of 8GB, but that was just the front screen in processes, not the private working set.

Kuca: I man a feature request for the wan accelerator yes, not the linux part.

[dan_il]

I've got a backup server that has recently really started struggling with a backup copy job through our wan accelerator and it seems to be tied to RAM. It will run fine for a night or two, but the Wan Accel memory usage continues to go up on evert subsequent run until it maxes out the allocated memory on the VM and the entire VM starts crawling. If I force restart the server to cancel the job, its will run fine again for a few more days. Is there any way to cap the amount of memory the Wan Accelerator is allowed to use so it stops getting hungrier and eating up all available RAM over time?

[MichaelCade]

Have you opened a support case?

I don't believe capping the service / process is the right way to resolve your issues.

[PTide]

Hi,

How much RAM do you have on the VM in total? Also, how do you tell it's WAN accelerator consuming RAM? Are any other Veeam roles present on the machine?

Thanks

[blithespirit]

What else does the WAN Accelerator host do, WA at a minimum will consume 8GB and anything else it finds, if you have other resource requirements then maybe the other services are being starved adding to the host issue. Min specs for WA is 4CPU and 8GB. Which WA is the issue, source or target, is it a one to one or many to one if target? Is the WA on your VBR server or a separate vm?