Compaq ProLiant 7000 Deploying the ProLiant Cluster for NetWare 4.2 - Page 27

'HSOR\LQJ WKH 3UR/LDQW &OXVWHU IRU 1HW:DUH   This state could be very damaging because, at this point, both servers have assumed control of the same shared drives. NHAS handles this situation by immediately recognizing this scenario and updating both servers to the true state of the cluster. As a result, as soon as Server A has resumed network communication, NHAS forces Server A into recovery mode and dismounts Server A's volumes. When the failover process has completed, NHAS will initiate the restoration of Server A and place both servers in costandby mode. Compaq recommends that system integrators/administrators test their clustered configuration for this and other conditions to ensure proper configuration. You may check the NHAS log file for console screen messages to review the response of your cluster to any tests. The console messages are logged in the HASERVER.LOG, which can be found in the SYS:\HASERVER directory. NDS It is not necessary for servers in the cluster to contain NDS replicas as long as trusted replicas of this NDS tree exist on the network. If one server in the cluster contains a trusted NDS replica then it is recommended, for availability reasons, that both servers contain a replica. If the two servers in the cluster are the only servers in the NDS tree and one contains the master record of the NDS tree, the other server should contain an NDS replica to preserve high availability of NDS. It doesn't matter whether the servers in the cluster reside in the same NDS organizational unit, but its is recommended that both servers of the ProLiant cluster reside in the same tree. NHAS translates the directory and volume trustee rights for the files and directories on the volume as per the NDS tree. This is performed by the HASTRUST.NLM. Should Server A fail and its shared drive failover to Server B, the newly mounted volumes must be manually added using one of the Novell NDS administration utilities to reflect the changes. Using NWAdmin is the recommended method. Long name spaces, Mac, NFS, etc. on a shared volume will transfer during a failover, but you need to make sure that the name space support modules for all name spaces in the cluster are loaded on all servers in the cluster. Name space support modules have a .NAM file extension, for example, LONG.NAM. When a server fails, any volume or directory information written on the drives at the time of failure might be damaged and in need of repair before it can be mounted by the surviving server. To protect against possible down time, you must have NetWare automatically run Vrepair on damaged volumes. NetWare performs this task by default unless you have disabled this feature. To check the status of this feature, enter set automatically repair bad volumes at the server console. If enabled, ON is returned. If OFF is returned, you can follow the instructions on the screen to enable this option. Additionally, when a server fails, a volume may be left in an undetermined state known as out of sync. In order to resynchronize a volume, you should use the NetWare installation utility. Print Services NHAS is an active/active data failover environment so both servers can actively be used for file, print, or application services. Unfortunately, with NetWare 4.11/IntraNetWare and NetWare 4.2, Novell legacy printing will function with NHAS but print queues assigned to shared volumes will not transfer in a failover. Novell did allow print queues to be assigned to specific volumes, but with their legacy printing product, the print queues are still internally assigned to a server name and volume. With the server name tied to the print queue it is not possible to transfer queued print jobs to the surviving server. It is strongly recommended that NHAS customers use Novell Distributed Print Services (NDPS) which fully supports cluster failovers. 0096-0699-A