HP Cisco Nexus 5000 Cisco MDS 9000 Family Release Notes for Cisco MDS SAN-OS R - Page 51

Caveats Send documentation comments to [email protected] • CSCsl42571 Symptom: SNMP timeouts occur when a AAA user ages out. By design, a AAA user is aged out every hour on a switch for security reasons. If a large fabric is discovered using a AAA user and a Performance Monitoring (PM) collection is added for such a fabric, a number of SNMP requests (related to the discovery or PM statistics collection) could time out. When a user views the PM statistics charts (in the Performance tab in the web client), the charts are not seen as continuous. Workaround: There are two workarounds for this issue. One or both of these workarounds can be used to mitigate this issue. - Turn on Interpolation by clicking on the Interpolation check box under Admin->Configure->Collections in the web client. This will insure that the charts are continuous in the case of any occasional legitimate timeouts. - Use a non-AAA user (for example, use a local user on the switch) for a large fabric discovery and for Performance Monitoring. For provisioning and configuration through the Fabric Manager web client, the user can still be authenticated remotely using AAA. • CSCsl65951 Symptom: Using Fabric Manager Release 3.2(2), an error is displayed in the creation wizard. This occurs when an enclosure spans multiple fabrics and not all fabrics are managed and when the Data Migration Wizard is used to create a job with that enclosure as the existing storage (selecting all ports listed in that enclosure). Workaround: Put the ports you plan to use as the existing storage in the migration into a separate enclosure, and use that enclosure in the wizard selection. • CSCsm08837 Symptom: When an IVR-enabled MDS switch with an empty device alias database, attempts to join a fabric which has approximately 7000 device aliases, the device alias merge fails. In this situation, the following occurs: - During the merge process between local and remote switches, the remote device alias database is received on the local switch. The local switch validates those device aliases with SAP 110 (which is IVR). - Since all 7000 aliases could not be sent in a single MTS message, the aliases are fragmented into 5 messages. - While IVR requires approximately 20 seconds to process each fragment, effectively it takes around 100 seconds to process all 5 messages. - Because DDAS has a timeout of around 60 seconds, the merge is rejected. - The merge process is retried after few minutes and the process repeats. Then finally failed. Workaround: Enable device alias CFS distribution before enabling IVR. • CSCsm47252 Symptom: DMM jobs move to the Reset state and the following reason is displayed: Peer connection failure. In a Cisco DMM dual-fabric topology, the Storage Service Module (SSMs) in the two fabrics communicate with each other over IP by establishing a TCP connection. This connection is routed IP over FC to the local Supervisor and from the Supervisor it is switched over the IP mgmt interface. As a result, if there is a Supervisor switchover, the TCP connection may or may not survive the switchover. In the event that the TCP connection cannot be re-established in time, the DMM jobs in that SMM will move to the Reset state. Workaround: None. OL-14116-10 Cisco MDS 9000 Family Release Notes for Cisco MDS SAN-OS Release 3.3(2) 51