IBM Tivoli Storage Manager, Version 7.1

Preventing long-running, inactive NDMP connections from closing

To prevent firewalls from closing network data-management protocol (NDMP) connections that are long-running but inactive, you can enable Transmission Control Protocol (TCP) keepalive on the NDMP control connections.

About this task

The Tivoli® Storage Manager server initiates control connections to network-attached storage (NAS) devices during NDMP backup or restore operations. These control connections might remain open and inactive for an extended amount of time. For example, suppose that two NDMP operations are started for the same NAS device. The control connection for one NDMP operation might remain open but inactive if the operation requires a resource, for example, a tape drive or sequential volume, that is being used by the other NDMP operation.

Some firewall software is configured to automatically close network connections that are inactive for a specified length of time. If a firewall exists between a Tivoli Storage Manager server and a NAS device, it is possible that the firewall can close NDMP control connections unexpectedly and cause the NDMP operation to fail.

The Tivoli Storage Manager server provides a mechanism, TCP keepalive, that you can enable to prevent long-running, inactive connections from being closed. If TCP keepalive is enabled, small packets are sent across the network at predefined intervals to the connection partner.

Restriction: To prevent errors, do not enable TCP keepalive in certain types of environments. One example is environments that do not have firewalls between the Tivoli Storage Manager server and a NAS device. Another example is environments with firewalls that tolerate long-running, inactive connections. Enabling TCP keepalive in these types of environments can cause an idle connection to be inadvertently closed if the connection partner temporarily fails to respond to TCP keepalive packets.


Feedback