Oracle Fabric Interconnect F1-15
 
Oracle Fabric Interconnect F1-4
 

PLA-Support>Sun Systems>SAND>Network>SN-SND: Oracle Virtual Networking
 

In this Document

Purpose

Details

Applies to:

Purpose

High ping latency may indicate need to use the 'allowed-vlans' option

Details

During the times of high ping latency it was observed in Fabric Interconnect logs, a high number of ibdma_no_ticket_errors.  Further analysis revealed a high amount of multicast traffic being sent from somewhere on the network.  The amount of multicast traffic was observed to be around 3,000 packets /second replicated to 128 vnics. We also observed ‘rnr packets’ increasing and MTAGs  at times rising above the 150 threshold.

Support asked for traffic  traces but in light of the fact that the high ping latency caused the customer applications to stall, Xsigo Support asked If customer if they would mind if we changed ethernet-port 4/1 from allowing traffic from all VLANS to pass through this IO Card by changing the ethernet-port ‘allowed-vlans=’ parameter,  to allow just the VLANs used by their environment.  

Support received permission from the customer  to make the changes listed below.  Please note, adding an allowed VLAN to the VLAN ACL takes 15 seconds to update the tables, so it is best to set the ethernet-port down prior to creating the VLAN ACL to not cause any outage.

*** An audit was performed of which VLANs were in use and a decision was made to change ‘allowed-vlans all’ to ‘allowed-vlans=310,320,335,336,337,339,689’.  The iocard 4 was set down, a removal of ‘allowed –vlans all’ was performed the, VLAN ACL was created and then the ethernet-port was set back up.  

Here are the commands that were run to create the allowed VLAN ACL on both IO Directors:

Once Support created the VLANs ACL, the customer reported the latency issue disappeared.

This solution has no attachment