2014-08-26 11:46 PM
We are using NetApp Cmode filer (CDOT), and have an synchronous FPolicy server, which responds to file access.
When large number of files are being accessed over CIFS/SMB share, we are observing problems at CIFS/SMB client end and the FPolicy server.
The behaviour is as mentioned below:
1.If the FPolicy server is under heavy load receiving screen request, we are not receiving keep_alive request, is this the expected behavior.
- verified that the external engine is connected.
- This behavior was observer after the netapp Vserver disconnects from the fpolicy server and then reconnects.
2. The parameter “max-server-reqs” is defined as
Maximum outstanding requests on the FPolicy server
Specifies the maximum number of outstanding requests that can be queued on the FPolicy server.
The range for this value is 1 through 10000. The default is 50.
- how do you define "outstanding request" Is it, the request that has been sent to the Fpolicy server and the response for the request is not yet received. (considering, external engine type is synchronous). Does it also treats all the request including KEEP_ALIVE and SCREEN_REQ as same, i.e if there are 50 unserved SCREEN_REQ request, then would the next KEEP_ALIVE message be counted as 51st request or 1st request.
- if this limit is reached how does, NetApp filer responds, to the smb/cifs request coming to the vservers.