Services¶
The Chef server has a built in process supervisor, which ensures that all of the required services are in the appropriate state at any given time. The supervisor starts two processes per service.
Service Subcommands¶
This command has a built in process supervisor that ensures all of the required services are in the appropriate state at any given time. The supervisor starts two processes per service and provides the following subcommands for managing services: hup, int, kill, once, restart, service-list, start, status, stop, tail, and term.
hup¶
The hup subcommand is used to send a SIGHUP to all services. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl hup SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
int¶
The int subcommand is used to send a SIGINT to all services. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl int SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
kill¶
The kill subcommand is used to send a SIGKILL to all services. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl kill SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
once¶
The supervisor for the Chef server is configured to restart any service that fails, unless that service has been asked to change its state. The once subcommand is used to tell the supervisor to not attempt to restart any service that fails.
This command is useful when troubleshooting configuration errors that prevent a service from starting. Run the once subcommand followed by the status subcommand to look for services in a down state and/or to identify which services are in trouble. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl once SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
restart¶
The restart subcommand is used to restart all services enabled on the Chef server or to restart an individual service by specifying the name of that service in the command.
Warning
When running the Chef server in a high availability configuration, restarting all services may trigger failover.
This subcommand has the following syntax:
$ chef-server-ctl restart SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand. When a service is successfully restarted the output should be similar to:
$ ok: run: service_name: (pid 12345) 1s
service-list¶
The service-list subcommand is used to display a list of all available services. A service that is enabled is labeled with an asterisk (*).
This subcommand has the following syntax:
$ chef-server-ctl service-list
start¶
The start subcommand is used to start all services that are enabled in the Chef server. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl start SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand. When a service is successfully started the output should be similar to:
$ ok: run: service_name: (pid 12345) 1s
The supervisor for the Chef server is configured to wait seven seconds for a service to respond to a command from the supervisor. If you see output that references a timeout, it means that a signal has been sent to the process, but that the process has yet to actually comply. In general, processes that have timed out are not a big concern, unless they are failing to respond to the signals at all. If a process is not responding, use a command like the kill subcommand to stop the process, investigate the cause (if required), and then use the start subcommand to re-enable it.
status¶
The status subcommand is used to show the status of all services available to the Chef server. The results will vary based on the configuration of a given server. This subcommand has the following syntax:
$ chef-server-ctl status
and will return the status for all services. Status can be returned for individual services by specifying the name of the service as part of the command:
$ chef-server-ctl status SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
When service status is requested, the output should be similar to:
$ run: service_name: (pid 12345) 12345s; run: log: (pid 1234) 67890s
where
run:is the state of the service (run:ordown:)service_name:is the name of the service for which status is returned(pid 12345)is the process identifier12345sis the uptime of the service, in seconds
For example:
$ down: opscode-erchef: (pid 35546) 10s
By default, runit will restart services automatically when the services fail. Therefore, runit may report the status of a service as run: even when there is an issue with that service. When investigating why a particular service is not running as it should be, look for the services with the shortest uptimes. For example, the list below indicates that the opscode-erchef should be investigated further:
run: oc-id
run: opscode-chef: (pid 4327) 13671s; run: log: (pid 4326) 13671s
run: opscode-erchef: (pid 5383) 5s; run: log: (pid 4382) 13669s
run: opscode-expander: (pid 4078) 13694s; run: log: (pid 4077) 13694s
run: opscode-expander-reindexer: (pid 4130) 13692s; run: log: (pid 4114) 13692s
High Availability¶
On back-end servers in a high availability topology, Keepalived is used by the clustering service to determine whether a service should be running. If the status subcommand is run against any of these nodes, a few things change:
- On the back-end node that is currently the backup server, it is normal to see only one running process: Keepalived
- On the back-end node that is currently the master server, it is normal to see all services running. It is also normal to see some services in a down state if the server, on reboot, did not attempt to start the services because Keepalived determines which services are restarted based on the state of the cluster
A sample status line for a service that is running on the master server in a high availability topology:
run: opscode-solr4: (pid 25341) 239s, normally down; run: log: (pid 5700) 145308s
Log Files¶
A typical status line for a service that is running any of the Chef server front-end services is similar to the following:
run: name_of_service: (pid 1486) 7819s; run: log: (pid 1485) 7819s
where:
rundescribes the state in which the supervisor attempts to keep processes. This state is eitherrunordown. If a service is in adownstate, it should be stoppedname_of_serviceis the service name, for example:opscode-solr4(pid 1486) 7819s;is the process identifier followed by the amount of time (in seconds) the service has been runningrun: log: (pid 1485) 7819sis the log process. It is typical for a log process to have a longer run time than a service; this is because the supervisor does not need to restart the log process in order to connect the supervised process
If the service is down, the status line will appear similar to the following:
down: opscode-solr4: 3s, normally up; run: log: (pid 1485) 8526s
where
downindicates that the service is in a down state3s, normally up;indicates that the service is normally in a run state and that the supervisor would attempt to restart this service after a reboot
stop¶
The stop subcommand is used to stop all services enabled on the Chef server. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl stop SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand. When a service is successfully stopped the output should be similar to:
$ ok: diwb: service_name: 0s, normally up
For example:
$ chef-server-ctl stop
will return something similar to:
ok: down: nginx: 393s, normally up
ok: down: opscode-chef: 391s, normally up
ok: down: opscode-erchef: 391s, normally up
ok: down: opscode-expander: 390s, normally up
ok: down: opscode-expander-reindexer: 389s, normally up
ok: down: opscode-solr4: 389s, normally up
ok: down: postgresql: 388s, normally up
ok: down: rabbitmq: 388s, normally up
ok: down: redis_lb: 387s, normally up
tail¶
The tail subcommand is used to follow all of the Chef server logs for all services. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl tail SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
term¶
The term subcommand is used to send a SIGTERM to all services. This command can also be run for an individual service by specifying the name of the service in the command.
This subcommand has the following syntax:
$ chef-server-ctl term SERVICE_NAME
where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.
List of Services¶
The following services are part of the Chef server:
- bifrost
- bookshelf
- keepalived
- nginx
- opscode-erchef
- opscode-expander
- opscode-solr4
- postgresql
- rabbitmq
- redis-lb
bifrost¶
The oc_bifrost service ensures that every request to view or manage objects stored on the Chef server is authorized.
status¶
To view the status for the service:
$ chef-server-ctl status bifrost
to return something like:
$ run: bifrost: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart bifrost
to return something like:
$ ok: run: bifrost: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once bifrost
bookshelf¶
The bookshelf service is an Amazon Simple Storage Service (S3)-compatible service that is used to store cookbooks, including all of the files—recipes, templates, and so on—that are associated with each cookbook.
status¶
To view the status for the service:
$ chef-server-ctl status bookshelf
to return something like:
$ run: bookshelf: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart bookshelf
to return something like:
$ ok: run: bookshelf: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once bookshelf
keepalived¶
The keepalived service manages the virtual IP address (VIP) between the backend machines in a high availability topology that uses DRBD.
status¶
To view the status for the service:
$ chef-server-ctl status keepalived
to return something like:
$ run: keepalived: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart keepalived
to return something like:
$ ok: run: keepalived: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once keepalived
nginx¶
The nginx service is used to manage traffic to the Chef server, including virtual hosts for internal and external API request/response routing, external add-on request routing, and routing between front- and back-end components.
status¶
To view the status for the service:
$ chef-server-ctl status nginx
to return something like:
$ run: nginx: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart nginx
to return something like:
$ ok: run: nginx: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once nginx
opscode-erchef¶
The opscode-erchef service is an Erlang-based service that is used to handle Chef server API requests to the following areas within the Chef server:
- Cookbooks
- Data bags
- Environments
- Nodes
- Roles
- Sandboxes
- Search
status¶
To view the status for the service:
$ chef-server-ctl status opscode-erchef
to return something like:
$ run: opscode-erchefs: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart opscode-erchef
to return something like:
$ ok: run: opscode-erchef: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once opscode-erchef
opscode-expander¶
The opscode-expander service is used to process data (pulled from the rabbitmq service’s message queue) so that it can be properly indexed by the opscode-solr4 service.
status¶
To view the status for the service:
$ chef-server-ctl status opscode-expander
to return something like:
$ run: opscode-expander: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart opscode-expander
to return something like:
$ ok: run: opscode-expander: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once opscode-expander
opscode-solr4¶
The opscode-solr4 service is used to create the search indexes used for searching objects like nodes, data bags, and cookbooks. (This service ensures timely search results via the Chef server API; data that is used by the Chef platform is stored in PostgreSQL.)
status¶
To view the status for the service:
$ chef-server-ctl status opscode-solr
to return something like:
$ run: opscode-solr: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart opscode-solr
to return something like:
$ ok: run: opscode-solr: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once opscode-solr
postgresql¶
The postgresql service is used to store node, object, and user data.
status¶
To view the status for the service:
$ chef-server-ctl status postgresql
to return something like:
$ run: postgresql: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart postgresql
to return something like:
$ ok: run: postgresql: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once postgresqls
rabbitmq¶
The rabbitmq service is used to provide the message queue that is used by the Chef server to get search data to Apache Solr so that it can be indexed for search. When Chef Analytics is configured, the rabbitmq service is also used to send data from the Chef server to the Chef Analytics server.
status¶
To view the status for the service:
$ chef-server-ctl status rabbitmq
to return something like:
$ run: rabbitmq: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart rabbitmq
to return something like:
$ ok: run: rabbitmq: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once rabbitmq
redis¶
Key-value store used in conjunction with Nginx to route requests and populate request data used by the Chef server.
status¶
To view the status for the service:
$ chef-server-ctl status redis
to return something like:
$ run: redis: (pid 1234) 123456s; run: log: (pid 5678) 789012s
restart¶
To restart the service:
$ chef-server-ctl restart redis
to return something like:
$ ok: run: redis: (pid 1234) 1234s
run once¶
To run the service, but not restart it (if the service fails):
$ chef-server-ctl once redis