olar.config, olar.config.common - Online Addition and
Removal policy configuration files
/etc/olar.config - Member-specific automatic deallocation
policy configuration file
/etc/olar.config.common - Cluster-shared automatic deallocation
policy configuration file
The olar.config and olar.config.common files provide userdefinable
policy settings that control the behavior of the
Automatic Deallocation Facility. The Automatic Deallocation
Facility will attempt to take a component out of service
without user intervention in the event that the component
is identified as being suspect by the component
indictment process. See the OLAR_intro(5) reference page
for information on the component indictment process. The
Web-Based Enterprise Service (WEBES) 4.0 software suite
(or higher) is required to be installed to support component
indictment.
By setting specific variables in either the /etc/olar.config
or /etc/olar.config.common file, you can define the
conditions for when automatic deallocation is appropriate.
Additionally, you can turn off automatic deallocation
individually for each supported component class, for example
all central processing units (CPUs), if desired.
Changes made to this file take effect immediately. There
is no need to restart or reconfigure any associated daemons.
Automatic deallocation is currently supported on the
GS80/GS160/GS320 series systems. CPU modules and memory
pages are currently supported for automatic deallocation
on these systems.
By default automatic deallocation policy applies to all
members in a cluster environment. This is specified
through the cluster-wide file /etc/olar.config.common.
However, when operating in a cluster environment, clusterwide
policy variables can be overridden using the member
specific configuration file /etc/olar.config.
Specifically, these files are used as follows: On a standalone
system, policy variables in both /etc/olar.config
and /etc/olar.config.common are used to determine automatic
deallocation policy, with attribute values in
/etc/olar.config overriding values set in /etc/olar.config.common.
In a cluster environment, configuration variables
in the /etc/olar.config.common file are shared by
all cluster members. The /etc/olar.config file is defined
as a Context Dependent Symbolic Link (CDSL) and hence,
there is a distinct /etc/olar.config file for each member
in a cluster. The configuration variable settings in any
given /etc/olar.config file provides automatic deallocation
policy for that member only.
When the Automatic Deallocation Facility is invoked as the
result of a component's indictment, it will post the
results of its execution, including specific policy variable
evaluation, as one or more Event Management (EVM)
events. This not only provides an audit trail for this
automated process, but also allows user applications to
listen for (or subscribe to) these events if desired. All
Automatic Deallocation Facility EVM events have a prefix
of sys.unix.sysman.auto_deallocate. See the EVM(5) reference
page for general information on the EVM facility.
You may view the complete list of all possible events that
may be posted, including a description of each event, by
issuing the following command: # evmwatch -i -f '[name
sys.unix.sysman.auto_deallocate]' | evmshow -t "@name" -x
| more
POLICY VARIABLES [Toc] [Back]
The following discussion describes the policy variables
defined for automatic deallocation of a CPU. Note that all
policy variables are case insensitive: The cpu_deallocate_allow
attribute is used to define whether or not
automatic deallocation is allowed when a CPU is indicted.
If this attribute is left NULL or specified as FALSE, then
there will be no automatic deallocation attempt of hardware
components that belong to the "CPU" class when a CPU
is indicted. By default this value is set to NULL. No
other cpu_deallocate* policy variables will be evaluated
if this attribute is not set to 'TRUE'. The cpu_deallocate_start_time
attribute denotes the time (in 24 hour
format) beginning with and after which automatic deallocation
is allowed. This attribute is used in conjunction
with cpu_deallocate_end_time to specify a time window in
which automatic deallocation of indicted CPUs can take
place. If no start value is specified, a default value of
00:00 is used. The cpu_deallocate_end_time attribute
denotes the time (in 24 hour format) up to and including
when automatic deallocation is allowed. This attribute is
used in conjunction with cpu_deallocate_start_time to
specify a time window in which automatic deallocation of
indicted CPUs can take place.
The start and end times are allowed to cross a day
boundary.
If no end value is specified, a default value of
23:59 is used. The cpu_deallocate_probability
attribute refers to the single indicted probability
or list of indicted probabilities for which automatic
deallocation should occur for an indicted
CPU. See the OLAR_intro(5) reference page for additional
information on component indictment probabilities.
Probabilities could be any combination of the three
discrete values 'LOW', 'MEDIUM' and 'HIGH'. When
specifying multiple probabilities they must be
enclosed in curly braces ( { } ), and must be
delimited by a comma (,). If no value is specified
for this attribute, automatic deallocation will
only occur for components indicted with a 'HIGH'
probability. The cpu_deallocate_user_supplied_script
attribute describes the full path to a
user-supplied script that you may want to execute
prior to automatic deallocation of an indicted CPU.
If present, this script must: Have execute permission.
Be owned by root. Provide a zero return
status to indicate successful execution. A nonzero
return value will prevent the automatic deallocation
from proceeding.
When the user supplied script is invoked, it is
passed two parameters. The first parameter is the
CPU name of the indicted CPU. The second parameter
is the CPU hardware-id of the indicted CPU. Determines
whether automatic deallocation of a CPU
should occur if processes have been bound to run
specifically on the indicted CPU, or processes have
been bound to the Resource Affinity Domain (RAD)
that the CPU belongs to. If the value of this policy
variable is TRUE, the CPU is automatically
removed from use by the operating system even under
the following situations: Processes are bound to
run specifically on the CPU. Those bound processes
will suspend until the CPU is brought back to the
'online' state. Processes are bound to the
Resource Affinity Domain (RAD) and the last running
CPU in that RAD has been indicted. Those processes
that are bound to the RAD will suspend execution
until any of the CPUs that belong to the RAD
are brought back to the 'online' state.
Conversely, if this policy variable is set to FALSE
or left NULL, an indicted CPU will not be deallocated
either if processes are bound to the CPU or
if processes are bound to the RAD of which the
indicted CPU is the last running CPU.
Use caution when enabling this option, because
allowing certain applications to suspend may cause
unpredicatable behavior. For information on binding
processes to CPUs or to RADs see the runon(1) and
rad_bind_pid(3) reference pages.
The following discussion describes the policy variables
defined for automatic deallocation of a memory page, also
known as a page frame number (PFN): The pfn_deallocate_allow
attribute is used to define whether or not
automatic deallocation is allowed when a memory page is
indicted. If this attribute is left NULL or specified as
FALSE, then there will be no automatic deallocation of
memory pages when a memory page is indicted. By default
this value is set to TRUE. No other pfn_deallocate* policy
variables will be evaluated if this attribute is not set
to 'TRUE'. The pfn_deallocate_start_time attribute
denotes the time (in 24 hour format) beginning with and
after which automatic deallocation is allowed. This
attribute is used in conjunction with pfn_deallocate_end_time
to specify a time window in which automatic
deallocation of indicted memory pages can take place. If
no start value is specified, a default value of 00:00 is
used. The pfn_deallocate_end_time attribute denotes the
time (in 24 hour format) up to and including which automatic
deallocation is allowed. This attribute is used in
conjunction with pfn_deallocate_start_time to denote a
time window in which automatic deallocation of indicted
memory pages can take place. The start and end times are
allowed to cross a day boundary. If no end value is specified,
a default value of 23:59 is used. The pfn_deallocate_probability
attribute refers to the single indicted
probability or list of indicted probabilities for which
automatic deallocation should occur for an indicted memory
page.
Probabilities could be any combination of the three
discrete values 'LOW', 'MEDIUM', and 'HIGH '. When
specifying multiple probabilities they must be
enclosed in curly braces ( { } ), and must be
delimited by a comma (,). If no value is specified
for this attribute, automatic deallocation will
only occur for memory pages indicted with a 'HIGH'
probability. The pfn_deallocate_user_supplied_script
attribute defines a full path to a
user-supplied script that you may want to execute
prior to automatic deallocation of an indicted PFN.
The script is started with a parameter that can be
used in the script, the decimal value of the PFN.
If present, this script must: Have execute permission.
Be owned by root. Provide a zero return
status to indicate successful execution. A nonzero
return value will prevent the automatic deallocation
from proceeding.
The automatic deallocation facility is only supported on
the GS80, GS160, and GS320 series systems.
This example sets the policy variables cpu_deallocate_start_time
and cpu_deallocate_end_time, to define a
time window in which CPU automatic deallocations are
allowed in the event that a CPU module had been indicted.
cpu_deallocate_start_time=22:00
cpu_deallocate_end_time=05:00
Note that start and end times cross a day boundary.
This example sets the policy variable pfn_deallocate_probability
to allow a memory page deallocation
when the associate memory page has been
indicted with either a medium or high probability.
pfn_deallocate_probability={MEDIUM,HIGH} This example
sets the policy variable cpu_deallocate_user_supplied_script
to the full path name of
a user-supplied script to be executed each time a
CPU module is indicted.
cpu_deallocate_user_supplied_script=/usr/users/las/bin/myScript.sh
If there is an error in any of the policy values, the
automatic deallocation utility will not deallocate the
indicted component or memory page. If all policy variables
were evaluated successfully, but there was an error when
attempting to complete the automatic deallocation, an
associated EVM event will be posted detailing the error.
Commands: hwmgr(8), sysman_menu(8), sysman_station(8)
Others: OLAR_intro(5)
olar.config(4)
[ Back ] |
