The following instructions provide additional information for operating Novell Cluster Services.
For instructions on configuring an OES Linux server as an iSCSI initiator and connecting to an iSCSI target, go to
Accessing iSCSI Targets on NetWare Servers from Linux Initiators
in the
iSCSI 1.1.3 Administration Guide for NetWare 6.5.
If you are connecting to an iSCSI target that already has NSS partitions and pools created on it, you may not be able to access those NSS partitions and pools until you either reboot the Linux intitiator server or run the evms_activate command at the Linux server console. This is required for each Linux initiator server that will access the iSCSI target.
If necessary, install NetWare and Novell Cluster Services, including the latest Service Pack on the server using the same node name and IP address.
If your SAN is not configured, install Novell Cluster Services after configuring the SAN.
If the Cluster object for the server is still present, use ConsoleOne® to delete the object.
You can do this by going to the Cluster container, selecting the node in the right frame, and pressing Delete.
Run the Novell Cluster Services installation.
The node will assume its former identity.
Cluster maintenance mode lets you temporarily suspend the cluster heartbeat while hardware maintenance is being performed. This is useful if you want to reset or power down the LAN switch without bringing down cluster servers. See Section 4.3, Novell Cluster Services Console Commands for more information.
If the master server in the cluster goes down while the cluster is in cluster maintenance mode, you must enter cluster maintenance off on all remaining cluster servers to bring the cluster out of maintenance mode. This is only necessary if the master server in the cluster goes down. If the master server in the cluster is up, you can enter cluster maintenance off on one server in the cluster to bring the entire cluster out of maintenance mode.
If you need to power down or recycle your shared storage system, you should shut down Linux Cluster Servers prior to doing so.
If LAN connectivity is lost between a cluster node and the other nodes in the cluster, it is possible that the lost node will be automatically shut down by the other cluster nodes. This is normal cluster operating behavior, and it prevents the lost node from trying to load cluster resources because it cannot detect the other cluster nodes.
By default, cluster nodes are configured to reboot after an automatic shutdown. On certain occasions, you might want to prevent a downed cluster node from rebooting so you can troubleshoot problems. To do this, edit the opt/novell/ncs/bin/ldncs file and find the following line:
echo -n $TOLERANCE > /proc/sys/kernel/panic
Replace $TOLERANCE with 0 to cause the server to not automatically reboot after a shutdown. After editing the ldncs file, you must reboot the server to cause the change to take effect.
If, during the OES cluster installation and configuration, you choose Remote System on the NCS LDAP Configuration page and you have LDAP configured to point to a NetWare 6.0 or earlier NetWare server, the cluster configuration will fail. To work around this problem, you must edit the etc/openldap/ldap.conf file and either disable certificates ( TLS_REQCERT <level> line) or change the file that contains the certificates ( TLS_CACERT <filename> line). See the ldap.conf man page for more information.
If you want to remove a cluster node from one cluster and add it to another cluster, you must reconfigure the node.
Bring down the cluster node.
In eDirectory, delete the cluster node object from the cluster container.
In eDirectory, find the NCP Server object and then go to Properties for that object.
This is not the cluster node object in the cluster container. It is the server object for the server.
Click , then find and delete the NCS:NetWare Cluster attribute.
Reconfigure Novell Cluster Services by following the procedure outlined in Section 3.5.3, Installing Novell Cluster Services after the OES Installation.
If you are planning to work with shared-disk NSS Pools and Volumes, you must install a shared-disk cluster by entering a device name for the cluster Split Brain Detector (SBD) partition at cluster creation (new cluster) time. If you don't enter a device name, you won't be able to cluster enable NSS Pools.
If you did not create a cluster partition during the Novell Cluster Services installation, you can create one later using the SBDUTIL utility. You must have a shared disk system (a Storage Area Network or SAN) connected to your cluster nodes before attempting to create a cluster partition. See Section 3.3, Shared Disk System Requirements for more information.
Before creating an SBD partition, you should first make sure one does not already exist on your shared disk system. To do this, enter sbdutil -f at the server console of a cluster node.
If a cluster partition already exists, do not create another one. If a cluster partition does not exist, enter sbdutil -c -d device_name at the server console. Replace device_name with the name of the device where you want to create the cluster partition.
For example, you might enter something similar to the following:
sbdutil -c -d /dev/sda
See the man page for sbdutil for more information on how to use it.
After creating the SBD partition, you must edit the Cluster object in eDirectory and enable the Shared Disk Flag attribute. You must then save changes and reboot the cluster. To do this
Start iManager, click , then click .
Enter the Cluster object name, or browse and select it, then click .
Under Valued Attributes, click , then click .
Check the check box, then click .
Click to save changes, then reboot the cluster.
To achieve a greater level of fault tolerance, you can mirror SBD partitions. You must use the evmsgui utility to create and mirror SBD partitions. If an SBD partition was created either during the Novell Cluster Services installation or later using the sbdutil command, you must delete that partition prior to creating and mirroring SBD partitions using evmsgui. To see if an SBD partition already exists, enter sbdutil -f at the server console of a Linux cluster server. (See Step 9 for more information on SBD partitions.)
If an SBD partition was created during the Novell Cluster Services installation or later using the sbdutil command, delete it.
Enter cluster down at the server console of one cluster server.
This will cause all cluster servers to leave the cluster.
Delete the SBD partition.
You can use nssmu, evmsgui, or other utilities to delete the SBD partition.
To create an SBD partition using evmsgui:
At the Linux server console of a cluster server, enter evmsgui to start the evmsgui utility.
Click , then click .
Click , choose the , then click .
Select , then click .
Specify 8 MB as the size of the cluster partition, then choose as the partition type.
Enter the name of your cluster as the Label, then click .
If necessary, repeat the above steps to create a second SBD partition.
To mirror SBD partitions:
At the Linux server console of a cluster server, enter evmsgui to start the evmsgui utility.
Click .
Locate one SBD partition and right-click it.
Select , then click .
Reboot all cluster nodes.