Oracle® Clusterware Administration and Deployment Guide 11g Release 2 (11.2) Part Number E16794-09 |
|
|
PDF · Mobi · ePub |
This appendix is a reference for Oracle Clusterware resources. It includes descriptions and usage examples of all resource attributes and a detailed description and examples of action scripts.
This appendix includes the following topics:
This section lists and describes attributes used when you register applications as resources in Oracle Clusterware. Use these attributes with the crsctl add resource
command, as follows:
$ crsctl add resource resource_name -type resource_type {[-attr "attribute_name='attribute_value', attribute_name='attribute_value' , ..."] | [-file file_name]}
List attribute-value pairs in a comma-delimited list after the -attr
flag and enclose the value of each attribute in single quotation marks (''
). Some resource attributes you cannot configure and are read only.
Alternatively, you can create a text file that contains the attribute-value pairs. For example:
PLACEMENT=favored HOSTING_MEMBERS=node1 node2 node3 RESTART_ATTEMPTS@CARDINALITYID(1)=0 RESTART_ATTEMPTS@CARDINALITYID(2)=0 FAILURE_THRESHOLD@CARDINALITYID(1)=2 FAILURE_THRESHOLD@CARDINALITYID(2)=4 FAILURE_INTERVAL@CARDINALITYID(1)=300 FAILURE_INTERVAL@CARDINALITYID(2)=500 CHECK_INTERVAL=2 CARDINALITY=2
This section includes the following topics:
This section describes the following resource attributes that you can configure when registering an application as a resource in Oracle Clusterware:
Note:
Values for the all attributes must be in lowercase. Attribute names must be in uppercase.Defines the owner of a resource and the access privileges granted to various operating system users and groups. The resource owner defines the operating system user of the owner and its privileges. You configure this optional attribute when you create a resource. If you do not configure this attribute, then the value is based on the identity of the process creating a resource. You can change the value of the attribute if such a change is allowed based on the existing privileges of the resource.
In the string:
owner
: The operating system user that owns a resource and the user under which the action script or application-specific agent run, followed by the privileges of the owner.
pgrp
: The operating system group that is the primary group of the owner of a resource, followed by the privileges of members of the primary group.
other
: Followed by privileges of others.
r
: This access option is the read option and gives the ability to only see a resource, its state, and configuration
w
: This access option gives the ability to modify a resource's attributes and to delete the resource
x
: This access option gives the ability to start, stop, and relocate a resource
By default, the identity of the client that creates a resource is the owner
. Also by default, root
, and the user specified in owner
have full privileges. You can grant required operating system users and operating system groups their privileges by adding the following lines to the ACL
attribute:
user:user_name:rwx group:group_name:rwx
ACL=owner:user_1:rwx,pgrp:osdba:rwx,other::r-
An absolute file name that includes the path and file name of an action script. The agent specified in the AGENT_FILENAME
attribute calls the script specified in the ACTION_SCRIPT
attribute.
ACTION_SCRIPT=fully_qualified_path_to_action_script
When set to 1
, Oracle Clusterware uses this attribute to reevaluate the placement of a resource during addition or restart of a cluster server. For resources where PLACEMENT=favored
, Oracle Clusterware may relocate running resources if the resources run on a non-favored server when a favored one joins the cluster.
ACTIVE_PLACEMENT=1
A fully qualified file name of an agent program that a resource type uses to manage its resources. Every resource type must have an agent program to manage its resources. Resource types use agent programs by either specifying a value for this attribute or inheriting it from their base resource type. There are two script agents included with Oracle Clusterware 11g release 2 (11.2): application
and scriptagent
. Oracle Clusterware uses the application
script agent for resources of the deprecated application
resource type. The default value for this attribute is scriptagent
.
AGENT_FILENAME=%Grid_home%/bin/application
Indicates whether Oracle Clusterware automatically starts a resource after a cluster server restart. Valid AUTO_START
values are:
always
: Restarts the resource when the server restarts regardless of the state of the resource when the server stopped.
restore
: Restores the resource to the same state that it was in when the server stopped. Oracle Clusterware attempts to restart the resource if the value of TARGET
was ONLINE
before the server stopped.
never
: Oracle Clusterware never restarts the resource regardless of the state of the resource when the server stopped.
The number of servers on which a resource can run, simultaneously. This is the upper limit for resource cardinality.
CARDINALITY=1
The time interval, in seconds, between repeated executions of the check
action. Shorter intervals enable more frequent checks but also increase resource consumption if you use the script agent. Use an application-specific agent to reduce resource consumption.
CHECK_INTERVAL=60
The number of instances of a cluster resource that can run on a single server.
DEGREE=1
Enter a description of the resource you are adding.
DESCRIPTION=Apache Web server
Oracle Clusterware uses this attribute to manage the state of the resource. Oracle Clusterware does not attempt to manage a disabled (ENABLED=0
) resource either directly or because of a dependency to another resource. A disabled resource cannot be started but it can be stopped.
Oracle Clusterware does not actively monitor disabled resources, meaning that Oracle Clusterware does not check their state. However, when Oracle Clusterware starts, it does query the state of disabled resources.
ENABLED=1
The interval, in seconds, during which Oracle Clusterware applies the FAILURE_THRESHOLD
attribute. If the value is zero (0
), then tracking of failures is disabled.
FAILURE_INTERVAL=30
The number of failures detected within a specified FAILURE_INTERVAL
for a resource before Oracle Clusterware marks the resource as unavailable and no longer monitors it. If a resource fails the specified number of times, then Oracle Clusterware stops the resource. If the value is zero (0
), then tracking of failures is disabled. The maximum value is 20
.
FAILURE_THRESHOLD=3
A space-separated, ordered list of cluster server names that can host a resource. This attribute is required only when using administrator management, and when the value of the PLACEMENT
attribute is set to favored
or restricted
. When registering applications as Oracle Clusterware
resources, use the SERVER_POOLS
attribute, instead.
Note:
For resources ofapplication
type, Oracle Clusterware places servers listed in the HOSTING_MEMBERS
attribute in the Generic server pool.See Also:
"Understanding Server Pools" for more information about the Generic server poolTo obtain a list of candidate node names, run the olsnodes
command to display a list of your server names.
HOSTING_MEMBERS=server1 server2 server3
Oracle Clusterware interprets the value of this attribute along with that of the PLACEMENT
attribute. When the value of PLACEMENT
is balanced
, the value of LOAD
determines where best to place a resource. A nonnegative, numeric value that quantitatively represents how much server capacity an instance of a resource consumes relative to other resources. Oracle Clusterware attempts to place resources on servers with the least total load of running resources.
LOAD=1
A case-sensitive alphanumeric string that names the resource. Oracle recommends a naming convention that starts with an alphanumeric prefix, such as sky1
, and complete the name with an identifier to describe it. A resource name can contain any platform-supported characters except the exclamation point (!) and the tilde (~). A resource name cannot begin with a period (.) nor with the string ora.
NAME=myApache
Controls offline monitoring of a resource. The value represents the interval (in seconds) that Oracle Clusterware monitors a resource when its state is OFFLINE
. Monitoring is disabled if the value is 0
.
OFFLINE_CHECK_INTERVAL=30
Specifies how Oracle Clusterware selects a cluster server on which to start a resource. Valid values are balanced
, favored
, or restricted
.
If you set the PLACEMENT
attribute to favored
or restricted
, then you must also assign values to the SERVER_POOLS
and HOSTING_MEMBERS
attributes. If you set the value of the PLACEMENT
attribute to balanced
, then the HOSTING_MEMBERS
attribute is not required.
PLACEMENT=favored
The number of times that Oracle Clusterware attempts to restart a resource on the resource's current server before attempting to relocate it. A value of 1
indicates that Oracle Clusterware only attempts to restart the resource once on a server. A second failure causes Oracle Clusterware to attempt to relocate the resource. A value of 0
indicates that there is no attempt to restart but Oracle Clusterware always tries to fail the resource over to another server.
RESTART_ATTEMPTS=2
The maximum time (in seconds) for an action to run. Oracle Clusterware returns an error message if the action script does not complete within the time specified. The timeout applies to all actions (start
, stop
, check
, and clean
).
SCRIPT_TIMEOUT=45
This attribute creates an affinity between a resource and one or more server pools regarding placement, and is dependent on the value of the PLACEMENT
attribute.
If a resource can run on any server in a cluster, then use the default value, *
, unless the resource is a cluster_resource
type, in which case, the default value for the SERVER_POOLS
attribute is empty. Only cluster administrators can specify *
as the value for this attribute. Otherwise, you must specify in the SERVER_POOLS
attribute a space-separated list of the server pools to which a particular resource can belong.
Use the PLACEMENT
attribute with the SERVER_POOLS
attribute, as follows: If you set the value of the PLACEMENT
attribute to either restricted
or favored
, then you must also provide a value for the SERVER_POOLS
attribute when using policy management for the resource.
If the value for PLACEMENT
is set to balanced
, then the resource only runs in the Generic and Free pools, unless SERVER_POOLS=*
.
See Also:
"Understanding Server Pools" for more information about server pools and "Role-Separated Management" for more information about cluster administratorsSERVER_POOLS=pool1 pool2 pool3
Specifies a set of relationships that Oracle Clusterware considers when starting a resource. You can specify a space-separated list of dependencies on several resources and resource types on which a particular resource can depend.
START_DEPENDENCIES=dependency(resource_set) [dependency(resource_set)] [...]
In the preceding syntax example the variables are defined, as follows:
dependency
: Possible values are hard
, weak
, attraction
, pullup
, and dispersion
. You can specify each dependency only once, except for pullup
, which you can specify multiple times.
resource_set
: A comma-delimited list of resource entities—either individual resources or resource types—enclosed in parentheses ()
, in the form of res1[, res2[, ...]]
, upon which the resource you are configuring depends.
Each resource entity is defined, as follows:
[modifier1:[modifier2:]] {resource_name | type:resource_type}
In the preceding syntax example, resource_name
is the name of a specific resource and type:
resource_type
is the name of a specific resource type. The resource type must be preceded by type
and the type
modifier must be the last resource entity in the list.
Optionally, you can specify modifiers to further configure resource entity dependencies. You can modify each dependency by prefixing the following modifiers to the resource entity:
hard([intermediate:][global:]{
resource_name
|
type:
resource_type
})
: Specify a hard
start dependency for a resource when you want the resource to start only when a particular resource or resource of a particular type starts.
Use intermediate
to specify that Oracle Clusterware can start this resource if a resource on which it depends is in either the ONLINE
or INTERMEDIATE
state. If not specified, then resources must be in the ONLINE
state for Oracle Clusterware to start this resource.
Use global
to specify that resources are not required to reside on the same server as a condition to Oracle Clusterware starting this resource. If not specified, then resources must reside on the same server for Oracle Clusterware to start this resource.
If you specify the hard
dependency on a resource type for a resource, then the resource can start if any resource of that particular type is running.
Note:
Oracle recommends that resources withhard
start dependencies also have pullup
start dependencies.weak([concurrent:][global:][uniform:]{
resource_name
|
type:
resource_type
})
: Specify a weak
start dependency for a resource when you want that resource to start despite whether named resources are running, or not. An attempt to start this resource also attempts to start any resources on which this resource depends if they are not running.
Use concurrent
to specify that Oracle Clusterware can start a dependent resource while a resource on which it depends is in the process of starting. If not specified, then resources must complete startup before Oracle Clusterware can start the dependent resource.
Use global
to specify that resources are not required to reside on the same server as a condition to Oracle Clusterware starting the dependent resource.
Use uniform
to start all instances of the resource everywhere the resource can run. If you do not specify a modifier (the default), then the resource starts on the same server as the resource on which it depends.
If you specify the weak
start dependency on a resource type for a resource, then the resource can start if any resource of that particular type is running.
attraction([intermediate:]{
resource_name
|
type:
resource_type
})
: Use the attraction
start dependency when you want this resource to run on the same server with a particular named resource or any resource of a particular type.
Use intermediate
to specify that this resource is attracted to resource entities on which it depends that are in the INTERMEDIATE
state. If not specified, then resources must be in the ONLINE
state to attract the dependent resource.
If you specify the attraction
dependency on a resource type for a resource, then any resource of that particular type attracts the dependent resource.
pullup[:always]([intermediate:][global:]{
resource_name
|
type:
resource_type
})
: When you specify the pullup
start dependency for a resource, then this resource starts as a result of named resources starting.
Use the always
modifier for pullup
so that Oracle Clusterware starts this resource despite the value of its TARGET
attribute, whether that value is ONLINE
or OFFLINE
. Otherwise, if you do not specify the always
modifier, then Oracle Clusterware starts this resource only if the value of the TARGET
attribute is ONLINE
for this resource.
Use intermediate
to specify that Oracle Clusterware can start this resource if a resource on which it depends is in either the ONLINE
or INTERMEDIATE
state. If not specified, then resources must be in the ONLINE
state for Oracle Clusterware to start this resource.
Use global
to specify that resources on which this resource depends are not required to reside on the same server as a condition to Oracle Clusterware starting this resource. If not specified, then resources on which this resource depends must reside on the same server for Oracle Clusterware to start this resource.
If you specify the pullup
dependency on a resource type for a resource, then, when any resource of that particular type starts, Oracle Clusterware can start this resource.
Note:
Oracle recommends that resources withhard
start dependencies also have pullup
start dependencies.dispersion[:active]([intermediate:]{
resource_name
|
type:
resource_type
})
: Specify the dispersion
start dependency for a resource that you want to run on a server that is different from the named resources or resources of a particular type. Resources may still end up running on the same server, depending on availability of servers.
Use the active
modifier to configure the dispersion
dependency so that Oracle Clusterware attempts to relocate the dependent resource to another server if it is co-located with another resource and another server comes online. Oracle Clusterware does not relocate resources to newly available servers unless you specify the active
modifier.
Use intermediate
to specify that Oracle Clusterware can relocate the dependent resource if a resource is in either the ONLINE
or INTERMEDIATE
state. If not specified, then resources must be in the ONLINE
state for dispersion of the dependent resource to occur.
See Also:
"Start Dependencies" for more details about start dependenciesThe maximum time (in seconds) in which a start action can run. Oracle Clusterware returns an error message if the action does not complete within the time specified. If you do not specify a value for this attribute or you specify 0
seconds, then Oracle Clusterware uses the value of the SCRIPT_TIMEOUT
attribute.
START_TIMEOUT=30
Specifies a set of relationships that Oracle Clusterware considers when stopping a resource.
STOP_DEPENDENCIES=dependency(resource_set) [dependency(resource_set)] ...
In the preceding syntax example the variables are defined, as follows:
dependency
: The only possible value is hard
.
resource_set
: A comma-delimited list, in the form of res1[, res2 [,...]]
, of resource entities—either individual resources or resource types—upon which the resource you are configuring depends.
Each resource entity is defined, as follows:
[modifier1:[modifier2:][modifier3:]] resource_name | type:resource_type
In the preceding syntax example, resource_name
is the name of a specific resource and type:
resource_type
is the name of a specific resource type. The resource type must be preceded by type:
.
Optionally, you can specify modifiers to further configure resource entity dependencies. You can modify each dependency by prefixing the following modifiers to the resource entity:
hard([intermediate:][global:][shutdown:]{
resource_name
|
type:
resource_type
})
: Specify a hard
stop dependency for a resource that you want to stop when named resources or resources of a particular resource type stop.
Use intermediate
to specify that the dependent resource can remain in an ONLINE
state if a resource is in either the ONLINE
or INTERMEDIATE
state. If not specified, then Oracle Clusterware stops the dependent resource unless resources are in the ONLINE
state.
Use global
to specify that the dependent resource remains in an ONLINE
state if a resource is in an ONLINE
state on any node in the cluster. If not specified, then when resources residing on the same server go offline, Oracle Clusterware stops the dependent resource.
Use shutdown
to apply this dependency when the Oracle Clusterware stack is shut down. This is a convenient way to affect the order of stopping resources when stopping the stack, without having any affect on planned or unplanned events on the individual resources. This dependency, when used with the shutdown
modifier, does not go into effect if somebody stops the resource directly, but only when the stack is shut down.
See Also:
"Stop Dependencies" for more details about stop dependenciesThe maximum time (in seconds) in which a stop or clean action can run. Oracle Clusterware returns an error message if the action does not complete within the time specified. If you do not specify this attribute or if you specify 0
seconds, then Oracle Clusterware uses the value of the SCRIPT_TIMEOUT
attribute.
STOP_TIMEOUT=30
The type of resource indicated when you create a resource. This attribute is required when creating a resource.
crsctl add resource resource_name -type resource_type
See Also:
"Resource Types" for details of resource typesThe value for UPTIME_THRESHOLD
represents the length of time that a resource must be up before Oracle Clusterware considers the resource to be stable. By setting a value for the UPTIME_THRESHOLD
attribute, you can indicate the stability of a resource.
Enter values for this attribute as a number followed by a letter that represents seconds (s), minutes (m), hours (h), days (d), or weeks (w). For example, a value of 7h
represents an uptime threshold of seven hours.
After the time period you specify for UPTIME_THRESHOLD
elapses, Oracle Clusterware resets the value for RESTART_COUNT
to 0
. Oracle Clusterware can alert you when the value for RESTART_COUNT
reaches the value that you set for RESTART_ATTEMPTS
.
Note:
Oracle Clusterware writes an alert to the clusterware alert log file when the value forRESTART_COUNT
reaches the value that you set for RESTART_ATTEMPTS
.You can view these attributes when you run the crsctl status resource
command on a particular resource. Oracle Clusterware sets these attributes when you register resources.
This is an internally-managed attribute that you can view only when you run the crsctl status resource
command on an ora
resource. You cannot edit this attribute.
An internally managed, read-only attribute that describes what, if any, action the policy engine is currently executing on the resource. Possible values and their meanings are as follows:
STARTING
: The policy engine is currently starting the resource
STOPPING
: The policy engine is currently stopping the resource
CLEANING
: The policy engine is currently cleaning the resource
STABLE
: The policy engine is not currently executing any action on the resource
Note, however, that the resource can still be locked as part of some command.
For cluster_resource
-type resources, this is an internally managed, read-only attribute that contains the name of the server on which the last start action for the resource succeeded.
For local_resource
-type resources, this is the name of the server to which the resource instance is pinned.
An internally managed, read-only attribute that describes when the policy engine registers the current state of the resource. Note that this may either be the timestamp of when state of the resource changed or when the policy engine discovered the state, as occurs when crsd
restarts.
This is an internally-managed attribute that you can view only when you run the crsctl status resource
command on an ora
resource. You cannot edit this attribute.
The Oracle Clusterware daemon counts the number of attempts to restart a resource, starting from zero up to the value specified in the RESTART_ATTEMPTS
attribute.
An internally managed attribute that reflects the current state of the resource as reported by Oracle Clusterware. The state of a resource can be one of the following:
ONLINE
: The resource is online and resource monitoring is enabled (see CHECK_INTERVAL
)
OFFLINE
: The resource is offline and only offline resource monitoring is enabled, if configured (see OFFLINE_CHECK_INTERVAL
)
INTERMEDIATE
: The resource is either partially online or was known to be online before and subsequent attempts to determine its state have failed; resource monitoring is enabled (see CHECK_INTERVAL
)
UNKNOWN
: The resource is unmanageable and its current state is unknown; manual intervention is required to resume its operation; the resource is not monitored
This is an internally-managed attribute that you can view only when you run the crsctl status resource
command on an ora
resource. You cannot edit this attribute.
An internally managed, read-only attribute that contains details about the state of a resource.
The four resource states—ONLINE
, OFFLINE
, UNKNOWN
, and INTERMEDIATE
—may map to different resource-specific values, such as mounted, unmounted, and open. Resource agent developers can use the STATE_DETAILS
attribute to provide a more detailed description of this mapping, resource to the resource state.
Providing details is optional. If details are not provided, then Oracle Clusterware uses only the four possible resource states. Additionally, if the agent cannot provide these details (as may also happen to the value of the resource state), then Oracle Clusterware sets the value of this attribute to provide minimal details about why the resource is in its current state.
An internal, read-only attribute that describes the desired state of a resource. Using the crsctl start
resource_name
or crsctl stop
resource_name
commands, however, can affect the value of this attribute.
This section includes examples of third-party applications using script agents.
Example B-1 shows an action script that fails over the Apache Web server.
Example B-1 Apache Action Script
#!/bin/sh HTTPDCONFLOCATION=/etc/httpd/conf/httpd.conf WEBPAGECHECK=http://<MyVIP>:80/icons/apache_pb.gif case $1 in 'start') /usr/sbin/apachectl -k start -f $HTTPDCONFLOCATION RET=$? ;; sleep(10) ;; 'stop') /usr/sbin/apachectl -k stop RET=$? ;; 'clean') /usr/sbin/apachectl -k stop RET=$? ;; 'check') /usr/bin/wget -q --delete-after $WEBPAGECHECK RET=$? ;; *) RET=0 ;; esac # 0: success; 1 : error if [ $RET -eq 0 ]; then exit 0 else exit 1 fi
Example B-2 shows the xclock script, which is a simple action script using xclock
available as a default binary on all Linux and UNIX platforms.
Example B-2 xclock Action Script
#!/bin/bash # start/stop/check script for xclock example # To test this change BIN_DIR to the directory where xclock is based # and set the DISPLAY variable to a server within your network. BIN_DIR=/usr/X11R6/bin LOG_DIR=/tmp BIN_NAME=xclock DISPLAY=yourhost.domain.com:0.0 export DISPLAY exit_code=0 if [ ! -d $BIN_DIR ] then echo "start failed" exit 2 fi PID1=`ps -ef | grep $BIN_NAME | grep -v grep | grep -v xclock_app | awk '{ print $2 }'` case $1 in 'start') if [ "$PID1" != "" ] then status_p1="running" else if [ -x $BIN_DIR/$BIN_NAME ] then umask 002 ${BIN_DIR}/${BIN_NAME} & 2>${LOG_DIR}/${BIN_NAME}.log status_p1="started" else echo `basename $0`": $BIN_NAME: Executable not found" exit_code=1 fi fi echo "$BIN_NAME: $status_p1" exit $exit_code ;; 'stop') if [ "${PID1}" != "" ] then kill -9 ${PID1} && echo "$BIN_NAME daemon killed" else echo "$BIN_NAME: no running Process!" fi exit $exit_code ;; 'check') if [ "$PID1" != "" ] then echo "running" exit 0 else echo "not running" exit 1 fi ;;*) echo "Usage: "`basename $0`" {start|stop|check}" ;; esac
Example B-3 shows an example of a shell script for an agent to monitor a file. When the agent is started, it creates the file (which is specified through an attribute) and when it is stopped, it deletes the file. The CHECK action consists of merely checking if the file exists or not. The variables with the _CRS_
prefix are attribute values that are provided to the script in its environment.
Example B-3 Action Script Example
#!/bin/sh TOUCH=/bin/touch RM=/bin/rm PATH_NAME=/tmp/$_CRS_NAME # # These messages go into the CRSD agent log file. echo " ******* `date` ********** " echo "Action script '$_CRS_ACTION_SCRIPT' for resource[$_CRS_NAME] called for action $1" # case "$1" in 'start') echo "START entry point has been called.." echo "Creating the file: $PATH_NAME" $TOUCH $PATH_NAME exit 0 ;; 'stop') echo "STOP entry point has been called.." echo "Deleting the file: $PATH_NAME" $RM $PATH_NAME exit 0 ;; 'check') echo "CHECK entry point has been called.." if [ -e $PATH_NAME ]; then echo "Check -- SUCCESS" exit 0 else echo "Check -- FAILED" exit 1 fi ;; 'clean') echo "CLEAN entry point has been called.." echo "Deleting the file: $PATH_NAME" $RM -f $PATH_NAME exit 0 ;; esac