For the Cluster Edition, several parameters have been added to the dataserver utility for starting and configuring a cluster.
This is the complete syntax for the dataserver command:
dataserver
-u, --admin-name=sa/sso_name
--buildquorum=[force]
-a, --caps-file=filename
-F, --cluster-input=filename
--cluster-takeover
-L, --conn-config-file=[filename]
--create-cluster-id [=quorum]
-D, --default-db-size=size_spec
-e, --error-log=[filename]
-G, --event-log-server=logserv_name
-f, --forcebuild
-H, --ha-server
-h, --help=[{0|1|2|3}[,display_width]]
--instance=instance_name
-y, --key-password=[key_password]
-K, --keytab-file=filename
-N, --license-prop-file=filename
-z, --logical-page-size=page_size
-Z, --master-db-size=size_spec
-d, --master-dev=master_device_name
-b, --master-dev-size=[size_spec]
-r, --master-mirror=filename
-m, --masterrecover
-g, --no-event-logging
-Q, --quorum-dev=quorum_dev
-q, --recover-quiesced
-w, --rewrite-db=database_name
-p, --sa-name={SSO_login_account | sso_role | sa_role}
-k, --server-principal=s_principal
-M, --shared-mem-dir=directory_name
-X, --sybmon
-T, --trace=trace_flag
-v, --version
--quorurm-dev=path_to_quorum_device – is the absolute path to the quorum device. It signals Adaptive Server to start as a shared-disk cluster. This is a required parameter for starting the Cluster Edition.
--cluster-input=input_file_name – loads the cluster configuration from the specified input file into the quorum device. Required when you include --buildquorum. You can use this parameter without buildquorum, but only when no other instances are running..
--instance=instance_name – name of the instance to start.
--create-cluster-id [=quorum] – pairs the master device and the quorum device. This parameter may be required in a disaster recovery situation. Use this parameter with caution because it overrides a safety feature that prevents the Cluster Edition from starting with the wrong combination of master and quorum devices.
The optional =quorum argument copies the existing identifier from the master device to the quorum device, writing the same unique ID to both quorum devices, allowing the master device to switch between them.
Be very careful when using the --create-cluster-id=quorum parameter.
The Adaptive Server algorithms that prevent you from starting an
instance that is already running, or two instances from simultaneously becoming
the coordinator, are based on the quorum device. The unique ID that
pairs the quorum device to the master device ensures that only one quorum
device can be used with that master. This unique ID protects you from
accidentially starting two clusters that use independent quorum devices
but the same master device. Using the --create-cluster-id=quorum parameter
weakens this protection by creating two quorums that use the same
master device. The system administrator must verify that only one quorum
is used at a time.
--buildquorum[=force] – creates a new quorum device. The force option overwrites an existing quorum device. This parameter requires that you include --cluster-input.
--cluster-takeover – used to perform a cluster takeover when you set automatic cluster takeover to zero (0). A cluster takeover is necessary when an abrubt total cluster failure leaves stale state information on the quorum device.
See the Utilities Guide for a complete description of the dataserver parameters.
Example 1 Create a new master and quorum device. The parameters in this example specify a 500 megabyte master device with a four kilobyte logical page size. The location of the master device, interface configuration file, interfaces path, and error log are specified in the cluster input file mycluster.inp. The name of the instance creating the quorum and master devices is "myase1":
dataserver --quorum-dev=/dev/raw/raw101 --buildquorum --cluster-input=mycluster.inp --logical-page-size=4K --master-dev-size=500M --instance=myase1
After the master database is built and the server is shut down, you can restart the instance with this:
dataserver --quorum-dev=/dev/raw/raw101 --instance=myase1
Example 2 Starts the instance myase2 as part of the cluster created in Example 1:
dataserver --quorum-dev=/dev/raw/raw101 --instance=myase2
Example 3 Recreates the quorum device for mycluster without overwriting the master device (starting the instance "myase1"). The cluster must be shutdown before rebuilding the quorum.:
dataserver --quorum-dev=/dev/raw/raw101 --buildquorum --cluster-input=mycluster.inp --instance=myase1
Reload the cluster configuration from the mycluster.inp input file without recreating the quorum device or the master device (the cluster must be shutdown first):
dataserver --quorum-dev=/dev/raw/raw101 --cluster-input=mycluster.inp --instance=myase1
The Cluster Edition pairs the master device with the quorum device to prevent you from accidentally starting your master device in multiple clusters. You can use --create-cluster-id to establish a relationship between an existing master device and an existing quorum device.
The dataserver command can read the following start options from the quorum device:
-d / --master-dev -e / --error-log -c / --config-file -i / --interface-dir -T / --trace
Options are honored in the following order:
Options that are passed to dataserver.
Options in the instance section of the cluster configuration file stored on the quorum device.
Options in the cluster section of the cluster configuration file stored on the quorum device.
In many cases you can start an instance by providing only the --quorum-dev and --instance parameters. However, any traceflags passed to dataserver with the -T or --trace parameters are combined with any traceflags stored on the quorum device.
dataserver does not honor the "additional run parameters" stored on the quorum device. These must be explicitly passed to dataserver (the values stored on the quorum device are read by the Unifed Agent and passed to dataserver when it is started by the agent).
Because Adaptive Server stores its passwords using a secure, one-way hash algorithm, you cannot recover forgotten passwords. The dataserver -p parameter generates new passwords for the system security officer login accounts and the system security officer (sso_role) and system administrator (sa_role) role accounts. Perform one of the following to reset the password:
Restart any instance with dataserver -p parameter.
Add a new instance with the dataserver -p parameter.
You must use the most recently generated password if more than one instance generates the password using the dataserver -p parameter. Once the Cluster Edition generates the password, reset the password to a more secure one.
