| PRIMECLUSTER Global Link Services Configuration and Administration Guide: Redundant Line Control Function 4.1 (for Solaris(TM) Operating System) |
|
Contents
Index
|
| Appendix D Notice of supplemental information | > D.2 Trouble shooting | > D.2.8 Resuming connection lags after switching (In the case of a Cluster system) |
When manually switching a node while communicating with other system, it takes time to resume the connection of IPv6 takeover address.
After performing takeover process of IPv6 between the nodes, the remote system primary connected to the node cannot instantly identify the Link-Layer Address for IPv6 takeover address. As a consequence, it takes approximately 30 seconds to enable a connection. To prevent such case, start in.ripngd (IPv6 routing daemon) beforehand in both operation and standby nodes to attain a connection instantly. A packet transmitted from the remote system is sent to Link-Layer Address of the primary node. This packet is being relayed from the primary node over to the new node while it reports the ICMP redirect message to the remote system. in.ripngd starts automatically with the system boot time by creating /etc/inet/ndpd.conf file. In such case, transfer path information of IPv6 is deployed to the adjacent router. In order to avoid deploying the transfer path information, after configuring an ip parameter by the following procedure, start in.ripngd with q option. Note, system reboot is not necessary to enable the following steps.
|
In the case of a Solaris 8 or Solaris 9: In the case of a Solaris 10: |
|
Contents
Index
|