SOFS: Event ID 30818 with SMB Direct for RDMA (KB 3052127)

One of my customers had an issue after implementing a Scale-Out File Server Cluster:

Issue:

They are running Windows Server 2012 R2 Systems, which are connected using InfiniBand. The Adapters are using SMB Direct between SOFS Node and Hyper-V Host to establish a Remote Direct Memory Access (RDMA) connection. If you restart a Hyper-V Host you will see Event ID 30818 logged:

Source: SMBClient
Event ID: 30818
Level: Warning
General: RDMA interfaces are available but the client failed to connect to the server over RDMA transport.
Server name: ServerName
Guidance:
Both client and server have RDMA (SMB Direct) adaptors but there was a problem with the connection and the client had to fall back to using TCP/IP SMB (non-RDMA).

You can see, the connection falls back top TCP/IP SMB and you can see this in decreasing Performance…

Cause:

In Windows Server 2012 R2 the SMB Direct Interface is started automatically by LanmanServer service. If this service starts and tries to open a RDMA connection, before SMB Direct could be loaded, the event is logged. The first connection is established via TCP/IP SMB. All new client connections are established using RDMA.

Solution:

My customer and the hardware vendor did not stop working on this issue (at this point my praise and respect – all in all it took 4 months), and now the solution could be found…its just a registry key:

Open Registry Editor, and then locate to the following registry subkey:

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\smbdirect

Right-click the Start registry entry, and then click Modify.
In the Value data box, change the value (by default, it is 3 as on-demand) to 2 (automatic).

This solution is now available for everyone in KB 3052127

So enjoy RDMA…

Dieser Post ist auch verfügbar auf: German