chef-backend-ctl

The Chef Infra Server backend HA cluster includes a command-line utility named chef-backend-ctl. This command-line tool is used to manage the Chef Infra Server backend HA cluster, start and stop individual services, and tail Chef Infra Server log files.

backup

Use the backup subcommand is to backup the data for a node in the backend HA cluster. This command is typically run against a follower node. Use the --force option to run this command against all nodes in the backend HA cluster. The backup is created as a tar.gz file and is located in /var/opt/chef-backup/.

Syntax

This subcommand has the following syntax:

chef-backend-ctl backup (options)

Options

This subcommand has the following options:

--with-external

Use to specify that data on an external PostgreSQL database should be removed.

Examples

Backup a node in the backend HA cluster

From a follower node, run the following command:

chef-backend-ctl backup

create-cluster

Use the create-cluster subcommand to initialize the cluster state, including the PostgreSQL data store, and then bootstrap the first node in a backend HA cluster or assist in the recovery of the entire backend HA cluster.

Syntax

This subcommand has the following syntax:

chef-backend-ctl create-cluster (options)

Options

This subcommand has the following options:

--elasticsearch-wait-time

The maximum amount of time (in seconds) to wait for Elasticsearch to start. Default value: 30.

--etcd-wait-time

The maximum amount of time (in seconds) to wait for etcd to start. Default value: 30.

--quorum-loss-recovery

Resets the cluster identifier in etcd to this node.

If nodes in a backend HA cluster are not available, etcd may not be able to form a cluster. If etcd cannot form a cluster, rebuild the cluster. First reset the cluster identifier on an active node, rebuild the nodes that will be part of the cluster, and then rejoin the rebuilt nodes to the cluster by using the chef-backend-ctl join-cluster subcommand.

-y, --yes

Do not prompt for confirmation.

Examples

None.

cleanse

The cleanse subcommand is used to re-set a machine in the Chef Infra Server backend HA cluster to the state it was in prior to the first time the reconfigure subcommand is run. This command will destroy all data, configuration files, and logs. The software that was put on-disk by the package installation will remain; re-run chef-backend-ctl reconfigure to recreate the default data and configuration files.

Syntax

This subcommand has the following syntax:

chef-backend-ctl cleanse

Options

This subcommand has the following options:

--with-external

Use to specify that data on an external PostgreSQL database should be removed.

Examples

None.

cluster-status

Use the cluster-status subcommand to return a list of all accessible nodes, their role (leader, follower), and the status for PostgreSQL and Elasticsearch.

Syntax

This subcommand has the following syntax:

chef-backend-ctl cluster-status (options)

Options

This subcommand has the following options:

--json

Return cluster health information as JSON.

Examples

Return cluster health data as JSON

chef-backend-ctl cluster-status --json

demote

Use the demote subcommand to demote the current leader in the backend HA cluster, after which a new leader is elected from the group of available followers in the backend HA cluster. This command will:

• Complete with an exit code of 0 if the original leader was demoted and a new leader was elected successfully.
• Return an error message and a non-zero exit code if leader election is prevented because failover has been disabled (for either the cluster or the node) or if a new leader could not be elected within the allowed time.

Syntax

This subcommand has the following syntax:

chef-backend-ctl demote

Examples

None.

force-leader

Use the force-leader subcommand to force the node from which the command is run to become the leader in the backend HA cluster if there is not already an elected leader.

This command should only be run if:

• The backend cluster appears to not have an elected and available leader
• All of the nodes in the backend HA cluster are believed to be in a healthy state; if one (or more) nodes are not healthy, first remove the unhealthy nodes, rebuild, and then rejoin them to the cluster

This command will:

• Run the chef-backend-ctl cluster-status subcommand to determine if a leader exists.

  Warning

  Nodes in the backend HA cluster may not be visible to each other when they are located in network partitions. This may prevent a cluster status check from discovering that a leader exists. If nodes exist on network partitions, to prevent this scenario, first run chef-backend-ctl cluster-status, and then verify the expected number of nodes in the backend HA cluster as healthy and waiting_for_leader before running this command.

• Complete with an exit code of 0 if the node from which the command is run becomes the leader.

• Return an error message and a non-zero exit code if a leader already exists.

Syntax

This subcommand has the following syntax:

chef-backend-ctl force-leader

Examples

None.

gather-logs

Use the gather-logs subcommand to gather the log files for a machine in the Chef Infra Server backend HA cluster into a tarball that contains all of the important log files and system information.

This subcommand has the following syntax:

chef-backend-ctl gather-logs

gen-sample-backend-config

Use the gen-sample-backend-config subcommand to generate output that contains all of the backend HA cluster settings along with their default values. Use this subcommand to get the values for publish_address and vip_interface prior to bootstrapping a new node for the backend HA cluster.

Syntax

This subcommand has the following syntax:

chef-backend-ctl gen-sample-backend-config

Example Output

The following example shows the results of running the chef-backend-ctl gen-sample-backend-config subcommand. The settings and output will vary, depending on the configuration. The elasticsearch, etcd, leaderl, and postgresql settings are generated automatically and should not be modified:

fqdn = 'be1'
hide_sensitive = true
ip_version = 'ipv4'
publish_address = '10.0.2.15'
vip = '10.0.2.15'
vip_interface = 'eth0'
etcd.client_port = 2379
etcd.enable = true
etcd.log_directory = '/var/log/chef-backend/etcd'
etcd.peer_port = 2380
etcd.log_rotation.file_maxbytes = 104857600
etcd.log_rotation.num_to_keep = 10
postgresql.archive_command = ''
postgresql.archive_mode = 'off'
postgresql.archive_timeout = 0
postgresql.checkpoint_completion_target = 0.5
postgresql.checkpoint_timeout = '5min'
postgresql.checkpoint_warning = '30s'
postgresql.data_dir = '/var/opt/chef-backend/postgresql/9.5/data'
postgresql.db_superuser = 'chef_pgsql'
postgresql.effective_cache_size = '496MB'
postgresql.enable = true
postgresql.hot_standby = 'on'
postgresql.keepalives_count = 2
postgresql.keepalives_idle = 60
postgresql.keepalives_interval = 15
postgresql.log_directory = '/var/log/chef-backend/postgresql/9.5'
postgresql.log_min_duration_statement = -1
postgresql.max_connections = 350
postgresql.max_replication_slots = 12
postgresql.max_wal_senders = 12
postgresql.max_wal_size = 64
postgresql.md5_auth_cidr_addresses = '["samehost", "samenet"]'
postgresql.min_wal_size = 5
postgresql.port = 5432
postgresql.replication_user = 'replicator'
postgresql.shared_buffers = '248MB'
postgresql.shmall = 4194304
postgresql.shmmax = 17179869184
postgresql.username = 'chef_pgsql'
postgresql.wal_keep_segments = 32
postgresql.wal_level = 'hot_standby'
postgresql.wal_log_hints = 'on'
postgresql.work_mem = '8MB'
postgresql.log_rotation.file_maxbytes = 104857600
postgresql.log_rotation.num_to_keep = 10
elasticsearch.data_dir = '/var/opt/chef-backend/elasticsearch/data'
elasticsearch.enable = true
elasticsearch.heap_size = 248
elasticsearch.jvm_opts = ''
elasticsearch.log_directory = '/var/log/chef-backend/elasticsearch'
elasticsearch.new_size = 32
elasticsearch.plugins_directory = '/var/opt/chef-backend/elasticsearch/plugins'
elasticsearch.port = 9200
elasticsearch.scripts_directory = '/var/opt/chef-backend/elasticsearch/scripts'
elasticsearch.temp_directory = '/var/opt/chef-backend/elasticsearch/'
elasticsearch.log_rotation.file_maxbytes = 104857600
elasticsearch.log_rotation.num_to_keep = 10
leaderl.control_worker_timeout_seconds = 30
leaderl.db_timeout = 2000
leaderl.enable = true
leaderl.health_check_interval_seconds = 2
leaderl.leader_ttl_seconds = 10
leaderl.log_directory = '/var/log/chef-backend/leaderl'
leaderl.status_internal_update_interval_seconds = 5
leaderl.status_post_update_interval_seconds = 10
leaderl.log_rotation.file_maxbytes = 104857600
leaderl.log_rotation.max_messages_per_second = 1000
leaderl.log_rotation.num_to_keep = 10
leaderl.etcd_pool.cull_interval_seconds = 60
leaderl.etcd_pool.http_timeout_ms = 5000
leaderl.etcd_pool.ibrowse_options = '{inactivity_timeout, infinity}'
leaderl.etcd_pool.init_count = 10
leaderl.etcd_pool.max_age_seconds = 60
leaderl.etcd_pool.max_connection_duration_seconds = 300
leaderl.etcd_pool.max_count = 10
ssl.certificate = nil
ssl.certificate_key = nil
ssl.ciphers = (a list of ciphers, not shown)
ssl.company_name = 'YouCorp'
ssl.country_name = 'US'
ssl.data_dir = '/var/opt/chef-backend/ssl/'
ssl.duration = 3650
ssl.key_length = 2048
ssl.organizational_unit_name = 'Operations'

gen-server-config

Use the gen-server-config subcommand to generate output for the /etc/opscode/chef-server.rb configuration file. This command may be run from any machine in the backend HA cluster, but must be run separately for each node that is part of the frontend group. This command will:

• Complete with an exit code of 0 if the chef-server.rb file is created successfully.
• Return an error message and a non-zero exit code if a node has not been bootstrapped or joined or if a FQDN is not provided.

Syntax

This subcommand has the following syntax:

chef-backend-ctl gen-server-config FQDN

Configure the Front End

1. On any node in the backend HA cluster, run the following command for each node in the frontend group:

   chef-backend-ctl gen-server-config FQDN -f chef-server.rb.fqdn
   

   where FQDN is the FQDN for the frontend machine. The generated chef-server.rb file will contain all of the values necessary for any frontend Chef Infra Server to connect to and bootstrap against the backend HA cluster.

2. On each frontend machine, install the chef-server-core package (version 12.4.0 or higher).

3. On each frontend machine, copy the generated chef-server.rb.fqdn to /etc/opscode/chef-server.rb.

4. On each frontend machine, with root permission, run the following command:

   chef-server-ctl reconfigure
   

Example Output

The following example shows the results of running the chef-backend-ctl gen-server-config subcommand. The settings and output will vary, depending on the configuration. These settings should be modified carefully:

fqdn "frontend1.chef-demo.com"
postgresql['external'] = true
postgresql['vip'] = '192.0.2.0'
postgresql['db_superuser'] = 'chef_pgsql'
postgresql['db_superuser_password'] = '...6810e52a01e562'
opscode_solr4['external'] = true
opscode_solr4['external_url'] = 'http://192.0.2.0:9200'
opscode_erchef['search_provider'] = 'elasticsearch'
opscode_erchef['search_queue_mode'] = 'batch'
bookshelf['storage_type'] = :sql
rabbitmq['enable'] = false
rabbitmq['management_enabled'] = false
rabbitmq['queue_length_monitor_enabled'] = false
opscode_expander['enable'] = false
dark_launch['actions'] = false
opscode_erchef['nginx_bookshelf_caching'] = :on
opscode_erchef['s3_url_expiry_window_size'] = '50%'

help

Use the help subcommand to print a list of all available chef-backend-ctl commands.

This subcommand has the following syntax:

chef-backend-ctl help

join-cluster

Use the join-cluster subcommand to configure a node to be a member of the backend HA cluster as a peer of the node at the specified PEER_NODE_IP IP address. This command will query the existing cluster to identify any necessary configuration information. If the --publish-address and --vip-interface options are not specified, and are specified in chef-backend.rb, the command will prompt with a list of items from which to choose.

This command will return an error message and a non-zero exit code when:

• The --secrets-file-path option is specified, a non-empty secrets file already exists at /etc/chef-backend/secrets.json, and the user declines to overwrite it.
• The --secrets-file-path option is specified, but does not specify a valid file and/or the file at /etc/chef-backend/secrets.json is empty or does not exist.
• The IP address specified by the --publish-address option does not exist on the node.
• The interface specified by the --vip-interface option does not exist on the node.
• Any IP address on the node is already registered in the backend HA cluster.
• The node is already configured for the backend HA cluster and the --recovery option is not specified.
• The --recovery option is specified, but no existing installation is found.
• --publish-address and/or --vip-interface are specified, but a non-empty chef-backend.rb file already exists. (This command will not overwrite a chef-backend.rb file.)

If successful, this command will generate a chef-backend.rb file at /etc/chef-backend/chef-backend.rb with the values for the publish_address, vip_interface, and vip added automatically.

Syntax

This subcommand has the following syntax:

chef-backend-ctl join-cluster PEER_NODE_IP (options)

where PEER_NODE_IP is the IP address of a peer in the cluster to be joined.

Options

This subcommand has the following options:

-i INTERFACE, --vip-interface INTERFACE

The network interface to which the backend VIP will bind in the event that this node becomes leader. If not provided, and if not specified in chef-backend.rb, this command will prompt to choose from a list of interfaces that are currently available on the node.

Note

This option should only be used the first time a node joins the backend HA cluster.

-p IP_ADDRESS, --publish-address IP_ADDRESS

The IP address that is published within the backend HA cluster. This IP address must be accessible to all nodes in the backend HA cluster. If not provided, and if not specified in chef-backend.rb, this command will prompt to choose from a list of IP addresses that are currently bound on the node.

Note

This option should only be used the first time a node joins the backend HA cluster.

--recovery

Force this node to rejoin the backend HA cluster if it has been removed via the chef-backend-ctl remove-node or chef-backend-ctl bootstrap --with-quorum-recovery commands.

Note

This option will run against the existing chef-backend.rb file, which means the --vip-interface and --publish-address options should not be specified. (They are already defined in the chef-backend.rb file.)

-s PATH, --secrets-file-path PATH

The ABSOLUTE path to the location of the secrets.json file on the bootstrapping node. The path should begin with a “/”. Default value: /etc/chef-backend/secrets.json.

-y, --yes

Do not prompt for confirmation.

Examples

None.

promote

Use the promote subcommand to promote the named node to be leader of the backend HA cluster. This command will:

• Complete with an exit code of 0 when the leader of the backend HA cluster is replaced as leader by the named node.
• Return an error message and a non-zero exit code if the named node is already leader because failover has been disabled (for either the cluster or the node) or if the new leader could not be promoted within the allowed time.

Syntax

This subcommand has the following syntax:

chef-backend-ctl promote NODE

Examples

None.

reconfigure

Use the reconfigure subcommand to reconfigure a machine in the Chef Infra Server backend HA cluster. This subcommand will also restart any services for which the service_name['enabled'] setting is set to true.

This subcommand has the following syntax:

chef-backend-ctl reconfigure

remove-node

Use the remove-node subcommand to remove the named node from the backend HA cluster by removing the node’s status from etcd and deleting it from the etcd cluster. This command is useful when a node is going to be replaced or if the join-cluster command was unsuccessful.

This command may not be run from the node that is to be removed; the node itself must be shut down physically or have all services stopped (via the the chef-backend-ctl stop command). If the node is still running or otherwise available to the backend HA cluster, this command will return an error message and a non-zero exist code.

Syntax

This subcommand has the following syntax:

chef-backend-ctl remove-node NODE_NAME

Options

This subcommand has the following options:

-y, --yes

Do not prompt for confirmation.

Examples

None.

restore

Use the restore subcommand to restore a backup created by the chef-backend-ctl backup subcommand. This command should be executed on the leader node in the backend HA cluster. This command will delete all existing data and replace it with the data in the backup archive.

Syntax

This subcommand has the following syntax:

chef-backend-ctl restore PATH (options)

where PATH is the path to a tar.gz file that was created by the chef-backend-ctl backup subcommand.

Options

This subcommand has the following options:

-d DIRECTORY, --staging-dir DIRECTORY

The path to an empty directory to be used during the restore operation. This directory must have enough available space to expand all data in the backup archive.

-y, --yes

Do not prompt for confirmation.

Examples

Restore data to the backend leader

From the leader node, run the following command:

chef-backend-ctl restore /var/opt/chef-backup/backup_file.tgz

set-cluster-failover

Use the set-cluster-failover subcommand to enable or disable failover across the backend HA cluster.

Syntax

This subcommand has the following syntax:

chef-backend-ctl set-cluster-failover STATE

where STATE may be one of on, off, true, false, enabled, or disabled.

set-node-failover

Use the set-node-failover subcommand to enable or disable failover for a node in the backend HA cluster.

Syntax

This subcommand has the following syntax:

chef-backend-ctl set-node-failover STATE

where STATE may be one of on, off, true, false, enabled, or disabled.

show-config

The show-config subcommand is used to view the configuration that will be generated by the reconfigure subcommand. This command is most useful in the early stages of a deployment to ensure that everything is built properly prior to installation.

This subcommand has the following syntax:

chef-backend-ctl show-config

status

Use the status subcommand to show the status of all services available to a node in the backend HA cluster. This subcommand has the following syntax:

chef-backend-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-backend-ctl status SERVICE_NAME

For example, full output:

chef-backend-ctl status

is similar to:

Service        Local Status        Time in State  Distributed Node Status
elasticsearch  running (pid 6661)  1d 5h 59m 41s  state: green; nodes online: 3/3
etcd           running (pid 6742)  1d 5h 59m 39s  health: green; healthy nodes: 3/3
leaderl        running (pid 6788)  1d 5h 59m 35s  leader: 1; waiting: 0; follower: 2; total: 3
postgresql     running (pid 6640)  1d 5h 59m 43s  leader: 1; offline: 0; syncing: 0; synced: 2

which shows status for a healthy backend HA cluster. The first three columns–Service, Local Status, and Time in State summarize the local state. The Distributed Node Status column shows:

• A three node cluster
• All nodes healthy and online
• A leader selected with two followers (for both leader/follower status for the leaderl service and a leader/synced state for the postgresql service)

Simple output:

chef-backend-ctl status --simple

is similar to:

run: elasticsearch: (pid 6661) 106983s; run: log: (pid 6667) 106983s
run: etcd: (pid 6742) 106981s; run: log: (pid 6630) 106984s
run: leaderl: (pid 6788) 106976s; run: log: (pid 6739) 106982s
run: postgresql: (pid 6640) 106984s; run: log: (pid 6653) 106983s

which shows the state of the services, process identifiers, and uptime (in seconds). Simple output is useful if one (or more) nodes in the backend HA cluster are down or in a degraded state.

uninstall

The uninstall subcommand is used to remove the Chef Infra Server application from a machine in the backend HA cluster, but without removing any of the data. This subcommand will shut down all services (including the runit process supervisor).

This subcommand has the following syntax:

chef-backend-ctl uninstall

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 for each service and provides the following subcommands for managing services: hup, int, kill, once, restart, service-list, start, status, stop, tail, and term.

hup

Use the hup subcommand to send a SIGHUP to all services on a machine in the Chef Infra Server backend HA cluster. 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-backend-ctl hup SERVICE_NAME

where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.

int

Use the int subcommand to send a SIGINT to all services on a machine in the Chef Infra Server backend HA cluster. 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-backend-ctl int SERVICE_NAME

where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.

kill

Use the kill subcommand to send a SIGKILL to all services on a machine in the Chef Infra Server backend HA cluster. 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-backend-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 a machine in the Chef Infra Server backend HA cluster 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-backend-ctl once SERVICE_NAME

where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.

restart

Use the restart subcommand to restart all services enabled on a machine in the Chef Infra Server backend HA cluster, or to restart an individual service by specifying the name of that service in the command.

This subcommand has the following syntax:

chef-backend-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

Use the service-list subcommand to display a list of all available services on a machine in the Chef Infra Server backend HA cluster. A service that is enabled is labeled with an asterisk (*).

This subcommand has the following syntax:

chef-backend-ctl service-list

start

Use the start subcommand to start all services that are enabled on a machine in the Chef Infra Server backend HA cluster. 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-backend-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 a machine in the Chef Infra Server backend HA cluster 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.

stop

Use the stop subcommand to stop all services enabled on the Chef Infra Server backend HA cluster. 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-backend-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-backend-ctl stop

will return something similar to:

ok: down: etcd: 393s, normally up
ok: down: postgresql: 388s, normally up

tail

Use the tail subcommand to follow all of the logs for all services on a machine in the Chef Infra Server backend HA cluster. 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-backend-ctl tail SERVICE_NAME

where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.

term

Use the term subcommand to send a SIGTERM to all services on a machine in the Chef Infra Server backend HA cluster. 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-backend-ctl term SERVICE_NAME

where SERVICE_NAME represents the name of any service that is listed after running the service-list subcommand.