ProxySQL

Supported OS Linux Mac OS Windows

Integration version7.0.0

Overview

This check monitors ProxySQL through the Datadog Agent.

Setup

Follow the instructions below to install and configure this check for an Agent running on a host. For containerized environments, see the Autodiscovery Integration Templates for guidance on applying these instructions.

Installation

The ProxySQL integration is included in the Datadog Agent package, so you don’t need to install anything else on your servers.

Configuration

Enabling SSL

To connect to ProxySQL using full SSL/TLS validation, enable the tls_verify option in conf.yaml. Include certificates and passwords needed to connect with SSL/TLS.

    tls_verify: true
    tls_ca_cert: ca_cert.pem

Host

To configure this check for an Agent running on a host:

  1. Edit the proxysql.d/conf.yaml file, in the conf.d/ folder at the root of your Agent’s configuration directory to start collecting your ProxySQL performance data. See the sample proxysql.d/conf.yaml for all available configuration options.

  2. Restart the Agent.

Log collection
  1. Collecting logs is disabled by default in the Datadog Agent, enable it in your datadog.yaml file:

    logs_enabled: true
    
  2. Add the log files you are interested in to your proxysql.d/conf.yaml file to start collecting your ProxySQL logs:

      logs:
          # Default logging file
        - type: file
          path: /var/log/proxysql.log
          source: proxysql
          service: "<SERVICE_NAME>"
          # Logged queries, file needs to be in JSON
          # https://github.com/sysown/proxysql/wiki/Query-Logging
        - type: file
          path: "<QUERY_LOGGING_FILE_PATH>"
          source: proxysql
          service: "<SERVICE_NAME>"
          # Audit log
          # https://github.com/sysown/proxysql/wiki/Audit-log
        - type: file
          path: "<AUDIT_LOG_FILE_PATH>"
          source: proxysql
          service: "<SERVICE_NAME>"
    

    Change the path and service parameter values and configure them for your environment. See the sample proxysql.d/conf.yaml for all available configuration options.

  3. Restart the Agent.

Containerized

For containerized environments, see the Autodiscovery Integration Templates for guidance on applying the parameters below.

Metric collection

ParameterValue
<INTEGRATION_NAME>proxysql
<INIT_CONFIG>blank or {}
<INSTANCE_CONFIG>{"host": "%%host%%", "port": "%%port%%", "username": "<USER>", "password": "<PASSWORD>"}
Log collection

Collecting logs is disabled by default in the Datadog Agent. To enable it, see Kubernetes Log Collection.

ParameterValue
<LOG_CONFIG>{"source": "proxysql", "service": "<SERVICE_NAME>"}

Validation

Run the Agent’s status subcommand and look for proxysql under the Checks section.

Data Collected

Metrics

proxysql.active_transactions
(gauge)
The number of client connections currently processing a transaction.
Shown as transaction
proxysql.backend.query_time_pct
(gauge)
Time spent (as a percentage of total time) making network calls to communicate with the backends.
Shown as percent
proxysql.backends.count
(gauge)
Number of mysql servers connected.
Shown as node
proxysql.client.connections_aborted
(gauge)
Client failed or improperly closed connections per second.
Shown as connection
proxysql.client.connections_connected
(gauge)
Client connections that are currently connected.
Shown as connection
proxysql.client.connections_created
(gauge)
The number of client connections created per second.
Shown as connection
proxysql.client.connections_non_idle
(gauge)
The number of client connections that are currently handled by the main worker threads.
Shown as connection
proxysql.client.statements.active_total
(gauge)
The number of prepared statements that are in use by clients.
Shown as unit
proxysql.client.statements.active_unique
(gauge)
The number of unique prepared statements currently in use by clients.
Shown as unit
proxysql.frontend.user_connections
(gauge)
Current number of frontend connections by user.
Shown as connection
proxysql.frontend.user_max_connections
(gauge)
Max number of frontend connections the user can create (as defined in mysqlusers.maxconnections).
Shown as connection
proxysql.memory.auth_memory
(gauge)
Memory used by the authentication module to store user credentials and attributes.
Shown as byte
proxysql.memory.jemalloc_active
(gauge)
Bytes in pages allocated by the application.
Shown as byte
proxysql.memory.jemalloc_allocated
(gauge)
Bytes allocated by the application.
Shown as byte
proxysql.memory.jemalloc_mapped
(gauge)
Bytes in extents mapped by the allocator.
Shown as byte
proxysql.memory.jemalloc_metadata
(gauge)
Bytes dedicated to metadata.
Shown as byte
proxysql.memory.jemalloc_resident
(gauge)
Bytes in physically resident data pages mapped by the allocator.
Shown as byte
proxysql.memory.jemalloc_retained
(gauge)
Bytes allocated by the application.
Shown as byte
proxysql.memory.query_digest_memory
(gauge)
Memory used to store data related to statsmysqlquery_digest.
Shown as byte
proxysql.memory.sqlite3_memory_bytes
(gauge)
Memory used by the embedded SQLite.
Shown as byte
proxysql.memory.stack_memory_admin_threads
(gauge)
Memory used by the stack of the admin threads.
Shown as byte
proxysql.memory.stack_memory_cluster_threads
(gauge)
Memory used by the stack of the cluster threads.
Shown as byte
proxysql.memory.stack_memory_mysql_threads
(gauge)
Memory used by the stack of the MySQL threads.
Shown as byte
proxysql.mysql.backend_buffers_bytes
(gauge)
Memory use of buffers related to backend connections.
Shown as byte
proxysql.mysql.frontend_buffers_bytes
(gauge)
Memory use of buffers related to frontend connections (read/write buffers and other queues).
Shown as byte
proxysql.mysql.monitor_workers
(gauge)
The number of monitor threads.
Shown as worker
proxysql.mysql.session_internal_bytes
(gauge)
Other memory used by ProxySQL to handle MySQL Sessions.
Shown as byte
proxysql.mysql.thread_workers
(gauge)
The number of MySQL Thread workers i.e. 'mysql-threads'.
Shown as worker
proxysql.performance.command.cnt_100ms
(count)
The number of commands of a type which executed within 100 milliseconds., but more than 50 milliseconds
Shown as command
proxysql.performance.command.cnt_100us
(count)
The number of commands of a type which executed within 100 microseconds.
Shown as command
proxysql.performance.command.cnt_10ms
(count)
The number of commands of a type which executed within 10 milliseconds., but more than 5 milliseconds
Shown as command
proxysql.performance.command.cnt_10s
(count)
The number of commands of a type which executed within 10 seconds., but more than 5 seconds
Shown as command
proxysql.performance.command.cnt_1ms
(count)
The number of commands of a type which executed within 1 milliseconds., but more than 500 microseconds
Shown as command
proxysql.performance.command.cnt_1s
(count)
The number of commands of a type which executed within 1 second., but more than 500 milliseconds
Shown as command
proxysql.performance.command.cnt_500ms
(count)
The number of commands of a type which executed within 500 milliseconds., but more than 100 milliseconds
Shown as command
proxysql.performance.command.cnt_500us
(count)
The number of commands of a type which executed within 500 microseconds., but more than 100 microseconds
Shown as command
proxysql.performance.command.cnt_50ms
(count)
The number of commands of a type which executed within 50 milliseconds., but more than 10 milliseconds
Shown as command
proxysql.performance.command.cnt_5ms
(count)
The number of commands of a type which executed within 5 milliseconds., but more than 1 milliseconds
Shown as command
proxysql.performance.command.cnt_5s
(count)
The number of commands of a type which executed within 5 seconds., but more than 1 second
Shown as command
proxysql.performance.command.cnt_infs
(count)
The number of commands whose execution exceeded 10 seconds.
Shown as command
proxysql.performance.command.total_count
(count)
The number of commands of a type executed.
Shown as command
proxysql.performance.command.total_time_pct
(gauge)
Total time spent executing commands of a type as a percentage of total time.
Shown as percent
proxysql.pool.bytes_data_recv
(gauge)
The amount of data received from the backend.
Shown as byte
proxysql.pool.bytes_data_sent
(gauge)
The amount of data sent to the backend.
Shown as byte
proxysql.pool.conn_failure
(gauge)
The number of requests per second where a connection was not available in the connection pool and either: a new connection had to be created or no backend was available.
Shown as connection
proxysql.pool.conn_immediate
(gauge)
The number of connections per second that a MySQL Thread obtained from its own local connection pool cache.
Shown as connection
proxysql.pool.conn_success
(gauge)
The number of requests per second where a connection was already available in the connection pool.
Shown as connection
proxysql.pool.connections_error
(gauge)
How many connections aren't established successfully.
Shown as connection
proxysql.pool.connections_free
(gauge)
How many connections are currently free. They are kept open in order to minimize the time cost of sending a query to the backend server.
Shown as connection
proxysql.pool.connections_ok
(gauge)
How many connections are established successfully.
Shown as connection
proxysql.pool.connections_used
(gauge)
How many connections are currently used by ProxySQL for sending queries to the backend server.
Shown as connection
proxysql.pool.latency_ms
(gauge)
The current ping time as reported from Monitor.
Shown as millisecond
proxysql.pool.latency_us
(gauge)
The current ping time as reported from Monitor.
Shown as microsecond
proxysql.pool.memory_bytes
(gauge)
Memory used by the connection pool to store connections metadata.
Shown as byte
proxysql.pool.queries
(gauge)
The number of queries routed towards this particular backend server.
Shown as query
proxysql.query_cache.bytes_in
(gauge)
The number of bytes per second sent into the Query Cache.
Shown as byte
proxysql.query_cache.bytes_out
(gauge)
The number of bytes per second read from the Query Cache.
Shown as byte
proxysql.query_cache.entries
(gauge)
The number of entries currently stored in the query cache.
Shown as entry
proxysql.query_cache.get.count
(gauge)
The number of read requests per second to the Query Cache.
Shown as query
proxysql.query_cache.get_ok.count
(gauge)
The number of successful read requests per second to the Query Cache.
Shown as query
proxysql.query_cache.memory_bytes
(gauge)
Memory usage of the query cache.
Shown as byte
proxysql.query_cache.purged
(gauge)
The number of entries purged per second by the Query Cache due to TTL expiration.
Shown as entry
proxysql.query_cache.set.count
(gauge)
The number of write requests per second to the Query Cache.
Shown as query
proxysql.query_processor_time_pct
(gauge)
The time (as a percentage of total time) spent inside the query processor determining the action to take with the query (internal module).
Shown as percent
proxysql.query_rules.rule_hits
(gauge)
The number of times query rules matched traffic.
Shown as hit
proxysql.questions
(gauge)
The number of client requests / statements executed per second.
Shown as question
proxysql.server.connections_aborted
(gauge)
Backend failed or improperly closed connections per second.
Shown as connection
proxysql.server.connections_connected
(gauge)
Backend connections that are currently connected.
Shown as connection
proxysql.server.connections_created
(gauge)
The number of backend connections created per second.
Shown as connection
proxysql.server.statements.active_total
(gauge)
The number of prepared statements currently available across all backend connections.
Shown as unit
proxysql.server.statements.active_unique
(gauge)
The number of unique prepared statements currently available across all backend connections.
Shown as unit
proxysql.slow_queries
(gauge)
The number of queries per second with an execution time greater than 'mysql-longquerytime' milliseconds.
Shown as query
proxysql.statements.cached
(gauge)
The number of global prepared statements for which proxysql has metadata.
Shown as unit
proxysql.uptime
(gauge)
The total uptime of ProxySQL in seconds.
Shown as second

Events

The ProxySQL check does not include any events.

Service Checks

proxysql.can_connect
Returns CRITICAL if the Agent can’t connect to ProxySQL, otherwise returns OK.
Statuses: ok, critical

proxysql.backend.status
Returns CRITICAL if ProxySQL considers the backend host as SHUNNED or OFFLINE_HARD. Returns WARNING if the backend host is OFFLINE_SOFT. Returns OK otherwise.
Statuses: ok, warning, critical

Troubleshooting

Need help? Contact Datadog support.