Managing Kinetica Services

Typical maintenance of the system may require starting & stopping of Kinetica services.

Note

The Active Analytics Workbench service is managed separately through its own kml service.

A variety of service status checks are available for assessing the health of the system and troubleshooting.


Managing Database Services


Start Database Services

The following database services can be started via GAdmin or command-line:

  • Database
    • httpd
    • Query Service
    • Text Search
  • Reveal
  • Graph Service

Tip

All services can be started via KAgent. Consult Admin for details.

Database Services Startup

There are two methods of starting database services:

GUI Startup

If GAdmin is not running, start system management processes first.

  1. Log in to GAdmin
  2. Click Admin on the left menu
  3. Click Start
  4. On the Start Service? pop-up, click Start
  5. Click Continue when the database has started
Command-Line Startup

This method will automatically start system management processes, if they are not already running.

Run the following as the root user:

systemctl start gpudb

Verify that Kinetica is running by browsing to http://<yourhostname>:8080/gadmin

Stop Database Services

The following database services can be stopped via GAdmin or command-line:

  • Reveal
  • Graph Service
  • Database
    • httpd
    • Query Service
    • Text Search

Tip

All services can be stopped via KAgent. Consult Admin for details.

GUI Stop

  1. Log in to GAdmin
  2. Click Admin on the left menu
  3. Click Stop
  4. On the Stop Service? pop-up, click Stop
  5. Click Continue when the database has stopped

Command-Line Stop

Run the following as the root user:

systemctl stop gpudb

Managing System Processes


Start System Management Processes

If the cluster was installed via KAgent, the statistics/metrics process, configuration manager, and optional high-availability queues need to be started first.

Run the following as the root user.

  1. Start the statistics/metrics manager on the KAgent node (usually head node):

    systemctl start kinetica_stats
    
  2. Start the configuration manager, etcd, on every node running it:

    systemctl start kinetica-etcd
    
  3. If this is a cluster in a High-Availability ring, start RabbitMQ on every node running it:

    systemctl start gpudb-mq
    

To start all system management processes:

  • Host Manager
  • GAdmin

Run the following as the root user on every node:

systemctl start gpudb_host_manager

Stop System Management Processes

To stop database services and all system management processes:

  • GAdmin
  • Host Manager

Run the following as the root user on every node:

systemctl stop gpudb_host_manager

Note

This will, in turn, stop database services, as well

If the cluster was installed via KAgent, the statistics/metrics process, configuration manager, and optional high-availability queues need to be stopped also.

Run the following as the root user.

  1. If this is a cluster in a High-Availability ring, stop RabbitMQ on every node running it:

    systemctl stop gpudb-mq
    
  2. Stop the configuration manager, etcd, on every node running it:

    systemctl stop kinetica-etcd
    
  3. Stop the statistics/metrics manager on the KAgent node (usually head node):

    systemctl stop kinetica_stats
    

Managing All Services

For KAgent installations, the following sections detail how to manage system processes via CLI.


Start All Processes

Run the following commands as the root user.

Start the statistics/metrics manager on the KAgent node (usually head node):

systemctl start kinetica_stats

Start the configuration manager, etcd, on every node running it:

systemctl start kinetica-etcd

If this is a cluster in a High-Availability ring, start RabbitMQ on every node running it:

systemctl start gpudb-mq

Start the Host Manager on every node:

systemctl start gpudb_host_manager

Start the database services on the head node:

systemctl start gpudb

Stop All Processes

Run the following commands as the root user.

Stop the database services on the head node:

systemctl stop gpudb

Stop the Host Manager on every node:

systemctl stop gpudb_host_manager

If this is a cluster in a High-Availability ring, stop RabbitMQ on every node running it:

systemctl stop gpudb-mq

Stop the configuration manager, etcd, on every node running it:

systemctl stop kinetica-etcd

Stop the statistics/metrics manager on the KAgent node (usually head node):

systemctl stop kinetica_stats

Managing Individual Components

The /opt/gpudb/core/bin/gpudb script has several options available to assist in managing the individual Kinetica components.

Important

The /opt/gpudb/core/bin/gpudb script should always be run as the gpudb user

Options:

OptionDescription
<component>-startStarts the given component if it's not currently running
<component>-stopStop the given component if it's not stopped
<component>-restartStop the given component if it's not stopped then start it
<component>-statusPrints status information, including the process IDs it is using
<component>-pidsPrints the process IDs the component is using
<component>-enabledReturns 1 is the component is enabled and 0 otherwise. Used internally by the gpudb script
<component>-installedReturns 1 is the component is enabled and 0 otherwise. Used internally by the gpudb script

Available components:

ComponentDescription
host-managerHost Management Services
gpudbDatabase Service
graphGraph Service
httpdWeb Server
query-plannerSQL Service
revealReveal Analytic Desktop
statsStatistics Services, available via GAdmin
text-searchFull Text Search Services
tomcatGAdmin Web Application

For example, restarting the stats server:

/opt/gpudb/core/bin/gpudb stats-restart

Note

  • If external_text_search = true in the /opt/gpudb/core/etc/gpudb.conf file, then the text search server can be managed using the options below. If external_text_search = false, then the text search server will not be able to be managed individually.
  • The Host Manager service manages the HTTPD and ODBC components. It's possible to stop these components, but Host Manager will restart them immediately. Stopping Host Manager and attempting to start these components individually will not work.

System Status Checks

There are several means to check the status of system components:

Processes

To check the status of the database processes:

  • Host Manager
  • GAdmin
  • Stats Services
  • Database
    • httpd
    • Query Service
    • Text Search
  • Reveal
  • Graph Service

Run the following as the root or gpudb user:

service gpudb status

GAdmin

The Kinetica Administration Application (GAdmin) provides a GUI for monitoring various aspects of the system:

API

To determine whether the REST endpoint services are operating, the Python API can be invoked, as follows:

$ /opt/gpudb/bin/gpudb_python /opt/gpudb/kitools/gadmin_sim.py -u <username> -p <password> --table --summary
+-----------------+--------------------------------+----------------------+----------------------+-------+
|     Schema      |           Table/View           |       Records        |       Type ID        |  TTL  |
+=================+================================+======================+======================+=======+
| SYSTEM          | <ALL TABLES/VIEWS>             |                    1 |                      |       |
| SYSTEM          | ITER                           |                    1 |        UNSET_TYPE_ID |    -1 |
+-----------------+--------------------------------+----------------------+----------------------+-------+

+---------------------------+----------------------+
|        Object Type        |        Count         |
+===========================+======================+
| Schemas                   |                    1 |
| Tables & Views            |                    1 |
| Records                   |                    1 |
| Records + Track Elements  |                    1 |
+---------------------------+----------------------+

Statuses

The tables below list the various statuses you could experience in GAdmin or in logs for the system itself, the nodes and ranks employed by the system, or all three.

Important

It's likely you won't see many of the node and rank statuses in GAdmin as they are transient and only last for a few seconds during start up or shut down.

System

StatusDescription
initThe INIT signal has been received and the system is initializing
rebalancingThe system is rebalancing data due to the addition/subtraction of ranks
runningThe system is up and available for requests
shutdownThe system has shut down and is not available
startingThe system is in the process of starting up
stoppedThe system has stopped all requests and is preparing for shutdown
system_limitedSomething is interfering with the systems's ability to communicate between all ranks

Node

StatusDescription
enum_hardwareThe node is processing the hardware being used and checking the license
establishing_clusterThe node is ensuring its connection to the system and the other hosts in the cluster
fatal_init_errorAn error occurred while validating the gpudb.conf configuration file
initThe INIT signal has been received and the node is initializing
parsed_confThe node has parsed configuration files for any changes
postThe node has started and informed the system
readyThe node has successfully shutdown and is ready to be started
runningThe node is up and available for requests
shutdownThe node has shut down and is not available
shutting_downThe node is in the process of shutting down and will not take requests
startedThe node has been started and is nearly ready for requests
startingThe node is in the process of starting up
stoppingThe node is stopping all requests and preparing for shutdown
validating_clusterThe node and any other nodes in the cluster are being validated by the system

Rank

StatusDescription
enum_hardwareThe rank is processing the hardware being used and checking the license
fatal_init_errorAn error occurred while validating the gpudb.conf configuration file
initThe INIT signal has been received and the rank is initializing
initializedThe rank has been primed for start-up
loaded_dataThe rank has successfully loaded data from the persist directory(ies)
loading_dataThe rank is in the process of loading data from the persist directory(ies)
not_respondingThe rank is currently not responding to requests
parsed_confThe rank has parsed configuration files for any changes
postThe rank has started and informed the system
runningThe rank is up and available for requests
shutdownThe rank has shut down and is not available
shutting_downThe rank is in the process of shutting down and will not take requests
startThe rank has received a start signal
startedThe rank has been started and is nearly ready for requests
startingThe rank is in the process of starting up
syncingThe rank is in the process of syncing types, tables, records, etc.
terminatedThe rank has encountered an error and was terminated; the rank will often restart if possible