DevOps Code ClearCase configuration for High Availability

To support High Availability, you must configure DevOps Code ClearCase® in the following ways:

About this task

Startup script. On each cluster node, remove the /etc/init.d/clearcase startup script, since VCS will control when DevOps Code ClearCase starts and stops.

Alternate hostnames. DevOps Code ClearCase must recognize the virtual hostname used by the DevOps Code ClearCase VCS service group in order to determine what views and VOBs are local to the system. Decide what virtual IP address and hostname will be used by the service group, and add them to the DNS server. On each cluster node, add the virtual hostname to the /var/adm/devops/clearcase/config/alternate_hostnames file.

View and VOB Servers. All views and VOBs managed by the HA cluster must be created with the {-host, -gpath, -hpath} triple, using storage on an NFS mount point on a NAS device (or other shared storage). Make sure the NFS mount point can be reached by all nodes in the cluster. If a NAS device is used for shared storage, the view or VOB owner must have read-write access to the NAS volume.

Examples:

cleartool mkview -tag vb-view1 -host cchaview 
-hpath /net/lockbox/vol/groupfolders/HA/vbrown/vb-view1.vws 
-gpath /net/lockbox/vol/groupfolders/HA/vbrown/vb-view1.vws 
/net/lockbox/vol/groupfolders/HA/vbrown/vb-view1.vws 
cleartool mkvob -tag /var/tmp/vb-vob1 -nc -host cchaview 
-hpath  /net/lockbox/vol/groupfolders/HA/vbrown/vb-vob1.vbs 
-gpath /net/lockbox/vol/groupfolders/HA/vbrown/vb-vob1.vbs 
/net/lockbox/vol/groupfolders/HA/vbrown/vb-vob1.vbs

As an alternative to the triple-path specification of VOB and view data storage, you can also create a storage location:

ct mkstgloc -view -host qvml358 -hpath /net/qvml024/var/tmp/hapath
-gpath /net/qvml024/var/tmp/hapath hastgloc /net/qvml024/var/tmp/hapath
ct mkview -stgloc hastgloc -tag myview
mkvob -tag myvob -stgloc hastgloc

Registry. The ClearCase registry files must be moved to an NFS mount point on a NAS device (or other shared storage). Make sure the NFS mount point can be reached by all nodes in the cluster. The root account for all cluster nodes must have read-write access to the NAS device, if NAS is used for shared storage.

The following are the recommended steps to move the registry files:

  1. Login to the registry host as root.
  2. Stop ClearCase: /opt/devops/clearcase/etc/clearcase stop
  3. Copy the rgy directory to NAS: cp –rp /var/adm/devops/clearcase/rgy NFS_mount_point
  4. Rename the old rgy directory: mv /var/adm/devops/clearcase/rgy /var/adm/devops/clearcase/rgy.old
  5. Symlink the NFS mount point: ln –s NFS_mount_point /var/adm/devops/clearcase/rgy
  6. Start ClearCase: /opt/devops/clearcase/etc/clearcase start

The registry configuration files are now in the /var/adm/devops/clearcase/config directory.

For the current registry master, the /var/adm/devops/clearcase/config/rgy_hosts.conf file must contain either its own physical hostname, or the virtual hostname of the registry server; and the /var/adm/devops/clearcase/config/rgy_svr.conf file must contain the word master. Executing the cleartool setrgysvrtype –master command (as root) on a master registry host will create the proper rgy_hosts.conf and rgy_svr.conf files.

In the same manner as for the master registry host, create a symbolic link to the NFS mount point on the standby host:
  1. Login to the standby registry host as root.
  2. Stop ClearCase: /opt/devops/clearcase/etc/clearcase stop
  3. Copy the rgy directory to NAS: cp –rp /var/adm/devops/clearcase/rgy NFS_mount_point
  4. Rename the old rgy directory: mv /var/adm/devops/clearcase/rgy /var/adm/devops/clearcase/rgy.old
  5. Symlink the NFS mount point: ln –s NFS_mount_point /var/adm/devops/clearcase/rgy
  6. Start ClearCase: /opt/devops/clearcase/etc/clearcase start

For the standby registry host, the /var/adm/devops/clearcase/config/rgy_hosts.conf file must contain the virtual hostname of the registry server; and the /var/adm/devops/clearcase/config/rgy_svr.conf file must contain the word standby. Executing the cleartool setrgysvrtype –standby command (as root) on the master registry server will turn it into a standby registry host, as long as it had previously been a standby registry host.

ClearCase clients must use the virtual hostname provided by VCS for the registry server. Replace the physical hostname with the virtual hostname in the /var/adm/devops/clearcase/config/rgy_hosts.conf file on each client machine.

License. DevOps Code ClearCase license keys are stored in /var/adm/devops/clearcase/license.db. DevOps Code ClearCase licenses are tied to a host id, which is unique to the physical host. In order to support license server failover, a duplicate set of license keys must be created for each cluster node, and stored locally on that node. Customers wishing to support license server failover within a cluster will need to make arrangements with IBM® for duplicate licenses, and should contact their IBM sales representative to discuss their requirements.

The DevOps Code ClearCase client machine must use the virtual hostname for the license server. Replace the physical hostname with the virtual hostname in the /var/adm/devops/clearcase/config/license_host file on each client machine.

FLEXlm provides its own form of license server failover, separate from VCS. Customers wishing to support FLEXlm license server failover should contact Macrovision for details.