Displaying cluster info

  • scinstall -pv - Display release and version numbers
  • scrgadm -p - Display resource group info
  • scstat - Display cluster status
  • scconf -p - Display cluster configuration
  • sccheck - Check global mount points (nothing is returned if everything is OK)

Stopping and starting a cluster

  • scshutdown -g0 -y - shutdown the entire cluster
  • scswitch -S -h anode - shutdown one node on a cluster (scswitch command switches all resources on anode to other nodes in the cluster depending on the cluster configuration
  • boot -x - boot up node in non cluster mode, i.e. it won't join the cluster

Controlling global devices and cluster file systems

Update the global device name space if a new device is added (this can also be done with boot -r)

  • scgdevs

To register a VXVM disk group as a disk device group (execute this on the primary node)

  • scsetup → 3 → 1

Or

  • scconf -a -D type=vxvm,name=dg1,nodelist=node1:node2

To register a VXVM volume changes e.g. adding volumes or removing volumes (on primary)

  • scsetup → 3 → 2

Or

  • scconf -c -D name=dg1,sync

Unregistering and removing a disk device group from cluster control

  • scsetup → 3 → 3

Or

  • scconf -r -D name=dg1

Changing the properties of a disk group, e.g. failback options

  • scsetup → 3 → 6

Or

  • scswitch -z -D disk-device-group -h anode

Creating a global filesystem

  1. newfs /dev/vx/rdsk/dg1/avol
  2. mkdir -p /global/device-group/mountpoint (/global is a standard used to indicate it's a global filesystem)
  3. mount using the global option and ufs logging (unless SDS metatrans is in use)
  4. update /etc/vfstab including global and logging options

N.B. Do not nest mount points with global filesystems

  • scsetup → 3 → 5 - Remove a node from a VXVM group

Disabling a rsource in a resource group

  • scswitch -n -j resource-name - to disable the resource
  • scswitch -z -g resource-group -h nodename - will bring the resource online but with the resource disabled

Cluster quorum control

  • scconf -pv | grep -i quorum - check quorum status
  • scsetup → 1 - add or rmove quorum devices (you need to know the device ID, DID, of the disk for most quorum processes. This can be found using scdidadm -L)

Removing the last quorum device:

  • scconf -c -q installmode (put cluster into install mode)
  • scconf -r -q globaldev=device (where device is teh DID)

Put quorum device into maintenance state:

  • scconf -c -q globaldev=device maintstate

Adminstering cluster interconnects

  • scstat -W - check interconnect status
  • scsetup → 3 - to edit/remove cluster interconnects

Administering cluster public networks

Sun Cluster 3.0 uses NAFO (Network Address FailOver) (Sun Cluster 3.1 uses IPMP)

Creating a NAFO group

  1. One (and only one) of the NAFO group should have a /etc/hostname.adapter entry
  2. This should be mapped to an entry in /etc/hosts
  3. pnmset -c nafogroup -o create adapter1 adapter2
  4. pnmset -l (to check)

Deleting a NAFO group

  1. ensure no resource are using the group - scrgadm -pvv
  2. if they are, switch resource group - scswitch -z -gresource-group -h node
  3. pnmset -c nafogroup -o delete
  4. pnmstat -l

Adding or removing an adapter

Use the -o add or -o remove option to pnmset command

Switching over active adapter

  • pnmset -c nafogroup -o switch adapter (adapter becomes the active adapter)

Tuning NAFO

/etc/cluster/pnmparams contains the tunable values for NAFO. These are:

  • inactive_time - number of seconds between probes
  • ping_timeout - timeout value for pings
  • repeat_test - number of times to do ping test before failing adapter
  • slow_network - number of seconds between pings
  • warmup_time - number of seconds to wai after failover before fault monitoring resumes

Examples of managing resources

Adding a new application resouce:
 scrgadm -a -t SUNW.gds -g rg-name -j resource-name \
 -x Start_command="/path/to/start/command/stopcommand" \
 -x Stop_command="/path/to/stop/command/stopcommand" \
 -x Probe_command="/path/to/probe/command/probecommand" \
 -x stop_signal=15 -x failover_enabled=true -y Start_timeout=600 -y Stop_timeout=300 \
 -y Port_list="23/tcp" -y Probe_timeout=180
Changing an existing resource
  • scrgadm -c -g resource_name -x failover_enabled=false
Recreating a logical hostname
  • scswitch -n -j resource-name; scrgadm -r -j resource-name
  • scrgadm -a -L -j resource-name -g resource-group -l hostname -n naforgroupid
  • scswitch -Z -g reource-group

Setting up a storage resource (non-shared)

  1. Use scsetup to register te volume group
  2. Add details to /etc/vfstab
  3. Create resource group: scrgdam -a -g resource-group -h nodename
  4. Register resource (if not already done) - scrgadm -a -t SUNW.HAStoragePlus
  5. Create resource: scrgadm -a -g resource-group -j resource-name -t SUNWHAStoragePlus -x FilesystemMountPoints=/mountpoint
  6. scswitch -Z -g resourcegroup

Filesystem should now be mounted on the defined node.

Recent Changes