NFS read transfer size

[mmcmax]

Hi everyone,

I spent last day to search why my gigabit cross-cable network connection with my nfs datastore seems to doesn't work near it's maximum speed.

In the network performance of the VI client i see only 7000 KBs during a suspend operation of a virtual machine with 1 Gigabite vRAM.

looking in /var/log/vmkwarning i found a strange line:

WARNING: NFS: 1222: Server 192.168.xxx.xxx supports read transfer of 262144 truncating to 65536

I search in the community but i don't find this message or a related post to this warning.

Is this related to wsize or rsize options in nfs server? There is a way to increase these parameters (or the corresponding value of the warning message) on the ESX server to match NFS server?

thanks

[nilsam]

I can see this in my enviroment also. I have a Datadomain DD530 which we use to backup our VM`s to. I get acceptable speed from my ESX servers about 20MB/s. I run about 4 backup streams simultaneously and reach speeds upto 700 Mbit/s.

On my Windows 2003 backup server I tuned it to Datadomain`s specification and that included increase the TCP send and recieve windows to 262144. That increased my CIFS thoughput 40-50%.

I have also tried to mount the NFS datastore in Service console with command: "mount -t nfs -o rw,tcp serverip/folder /mount point" and got about twice the speed.

[gerrybrennan]

Hello,

I have my VMWare (esx) server on a local drive, however all of my VM storage is over NFS. I have the same error as mention before with Warning: NFS: 1222:, however I am getting this when I deploy a VM from a template. Is there a setting somewhere that I may be missing to disable / remove the truncating?

Thank you

Gerry