View Source

SMF consists of four command line utilities:
* *svcs:* allows you to examine the state of your services and determine what went wrong.
* *svcadm:* enable, disable, and restart a service.
* *svccfg:* load manifest files (XML) that maintain configurations for each service.
* *svcprop:* retrieves properties on a service (useful when writing custom scripts)

*In this topic:*

h1. {anchor:svcs}Examining Service Status - svcs

The {{svcs}} command displays information about the state of your services. This typically means whether or not they are running and any problems encountered when attempting to start them. In general, services will fall into three different states:
* _online_: the service is enabled and functioning normally
* _offline_: the service is stopped and disabled
* _maintenance_: the service has encountered a problem and is on hold until the problem is addressed by an administrator.

Running the {{svcs}} command with the {{\-a}} argument displays a list of all online and offline services. By default {{svcs}} prints out:
* _status_: the current state of the service.
* _stime_: when the service entered the current state.
* _FMRI_: the name of the service.

Each service is identified by a Fault Management Resource Identifier (FMRI). For example, the FMRI for the Apache service is:
The above FMRI breaks down in the following way:
|| String || Description ||
| {{svc:}} | The service type |
| {{/network/http}} | The service name |
| {{:cswapache2}} | The service instance |
For example, this output shows that Apache was disabled on May 31st:
$ svcs svc:/network/http:cswapache2
disabled May_31 svc:/network/http:cswapache2
An easy way of uncovering the FMRI for a service is by using {{grep}} in combination with {{svcs \-a}}. For example, using grep in the following exposes the FMRI of the MySQL service, which is {{cswmysql5}}:
$ svcs -a | grep -i mysql
enabled May_31 svc:/network/cswmysql5:default
You can abbreviate an FMRI by specifying the instance name or the trailing portion of the service name. For example, valid abbreviations for {{svc:/network/http:cswapache2}} are:
* {{cswapache2}}
* {{:cswapache2}}
* {{http}}
* {{http:cswapache2}}
* {{network/http{}}}{tip}

h1. {anchor:svcadm}Starting and Stoping Services - svcadm

The {{svcadm}} command is used to enable, disable, restart, or refresh services. For example, this command enables the MySQL service.
$ sudo svcadm enable cswmysql5
Using {{svcs}}, you can verify that the service is enabled:
$ svcs cswmysql5
online 15:20:39 svc:/network/cswmysql5:default
You use {{disable}} to stop services. For example, this stops the MySQL service:
$ sudo svcadm disable cswmysql5
You use {{restart}} to refresh a service. For example, after making a configuration change, you can refresh an enabled service like this:
$ sudo svcadm restart cswmysql5
Some daemons do not respond to the {{restart}} command. If that is the case, you will need to disable and re-enable the service.{note}

h2. Conducting Service Maintenance - svcadm clear

If a service enters maintenance mode or becomes disabled, you will need to perform some maintenance before SMF can restart the service. This means resolving any conflicts that prevent the service from running then using {{svcadm clear}} to clear the service state. When using {{svcadm clear}}, you need to specify the service FMRI.

This command does nothing to fix a service, it just signals that the service is ready to resume. Before you can clear a service state, you need to ensure that you resolve the conflict that caused the service to go into maintenance mode. For example, if you configure Postgres to use an IP port that is already in use by another service, the Postgres service will terminate abnormally and go into maintenance mode. In this case, you would need to resolve the IP conflict by modifying the Postgres configuration to use another IP or terminating the conflicting service. You would then clear the state after resolving the conflict.

Once the state is cleared, the service can resume. Each service has an assigned service restarter agent that is responsible for carrying out actions against it. The default service restarter is {{svc.startd}}.
* If a service instance is in maintenance mode, this command informs the restarter agent that the service was repaired. 
* If a service instance is disabled, this command requests that the restarter agent transition the service to the online state.

For example, the following will clear the state of the Apache service and SMF will automatically restart the service once cleared:
{code:lang=none}sudo svcadm clear svc:/network/http:cswapache2{code}

h2. Verifying a Service is in Maintenance Mode

SMF will place a service in maintenance mode when the service encounters something that causes it to crash. This usually indicates an error with the service but can also occur if your SmartMachine is running out of resources (RAM or disk space).

To verify if a service is in maintenance mode:
# Run this command:
svcs -a
This will show all running and disabled services. If a service is in maintenance mode, you will see something similar to this:
maintenence 18:50:25 svc:/network/webmin:webmin 
# Review the log to root-cause why the service was in maintenance mode.
# Take the service out of maintenance mode:
svcadm clear service_name

h2. Verifying a Service is Disabled

To verify if a service is disabled:
{code:lang=none}svcs -a{code}
If the service is disabled, you will see something similar to this:
disabled 18:51:10 svc:/network/webmin:webmin

h2. Enabling a Service

To enable a service:
# Enable the service:
svcadm enable service_name
# Verify the service is enabled:
svcs -a | grep service_name

If successful, you should see something similar to this:
online 18:50:25 svc:/network/webmin:webmin
The term "online" is the service state and indicates that the service is running.

h1. Examining Service Contracts - svcs \-p

SMF maintains a contract with every running service it manages. The contract keeps track of what processes are running for any given service. Using the {{\-p}} option, you can determine all the processes that belong to a service. In the following example, the MySQL daemon is process number 29004.
$ svcs -p network/cswmysql5
online 16:55:27 svc:/network/cswmysql5:default
16:55:27 28938 mysqld_safe
16:55:27 29004 mysqld
The following example demonstrates how SMF restarts a service when it stops unexpectedly:
$ kill -9 29004
$ svcs -p network/cswmysql5
online* 17:00:01 svc:/network/cswmysql5:default
16:55:27 28938 mysqld_safe
17:00:01 29228 mysqld
$ mysql -u mysql
mysql> \q
Even though the MySQL daemon was unexpectedly terminated, it was automatically restarted by SMF. Notice that the STIME shows that the MySQL service is back online. The inclusion of an asterisk with the "online" state indicates that the service is currently in transition. However, the MySQL service is already back online by the time the next command is run.

SMF does not restart the service in brain-dead mode like a legacy {{inittab}}; You can configure a threshold for service restarts. For example, numerous restarts of a service in a 60 second time frame might indicate a severe issue in your environment. You can configure a restart threshold in SMF for that service. At that point, SMF will put the service in "maintenance" mode, and the service will remain in that state until you clear it with {{svcadm clear}}.

h1. {anchor:svccfg}Configuring Services - svccfg

The {{svccfg}} command allows you to import, export, and modify service configurations. You specify entities to manipulate by using the {{\-s}} option with an FMRI. The following example will set an environment variable for the specified FMRI with the value you specify.
{code:lang=none}svccfg -s FMRI setenv ENV_VARIABLE value{code}
You can invoke {{svccfg}} directly with individual subcommands or by specifying a script file.{note}
If you make any changes to a service using this command, you need to restart the service for the changes to take effect.{note}

h1. Enabling SMF Access

if you want to enable users with no sudo access to manage SMF, you can modify a user profile as follows:
# Open {{/etc/user_attr}} for edit.
# Add this line replacing "myuser" with the login you want to enable:
{code:lang=none}myuser::::profiles=Service Management{code}

After this change, the specified user is able to manage SMF (import, stop, start) without access to {{sudo}}. This minimizes the need for unnecessarily sharing sudo access among users.

h1. Checking properties of a service

To retrieve a specific setting for a service, we use svcprop in this case to figure out which ipf.conf ipfilter is using:

# svcprop -p firewall_config_default/custom_policy_file network/ipfilter:default

h1. Uncovering Information About Services

Using SMF to uncover information about a service is easy.

|| Command || Description ||
| {{svcs \-a}} | List all services for this SmartMachine, including disabled services. |
| {{svcs \-x}} | List explanations for all services that are running but not enabled or services that are preventing another service from running. |
| {{svcs \-p NAME}} | List all processes for the service you specify. |

h1. References

* [HOWTO: Solaris SMF & Startup Scripts, How to be reboot ready|]
* [SMF System Administration Guide|]
* [The nice Facility - How to have a command run after a service is started by SMF|]