Observability

ElyraSQL exposes runtime health through MySQL-compatible introspection plus a slow-query log, so existing MySQL tooling works.

Server status

SHOW STATUS (also SHOW GLOBAL STATUS / SHOW SESSION STATUS) returns process-wide counters as Variable_name / Value rows:

SHOW STATUS;
SHOW STATUS LIKE 'Com_%';

Variable	Meaning
Uptime	Seconds since the server started
Threads_connected	Current open connections
Connections	Total connections since start
Questions / Queries	Total statements executed
Com_select / Com_insert / Com_update / Com_delete / Com_other	Per-type statement counts
Errors	Statements that returned an error
Slow_queries	Statements at or above the slow threshold

A trailing LIKE 'prefix%' filters by variable-name prefix.

Process list

SHOW PROCESSLIST (or SHOW FULL PROCESSLIST) lists live connections with the standard columns Id, User, Host, db, Command, Time, State, Info. A connection running a statement shows Command = Query, the elapsed Time in seconds, and the SQL in Info; an idle connection shows Command = Sleep.

SHOW PROCESSLIST;

Slow-query log

Start the server with a millisecond threshold to log statements that take at least that long:

elyrasql serve --data elyra.edb --slow-query-ms 500
# or
ELYRASQL_SLOW_QUERY_MS=500 elyrasql serve --data elyra.edb

Each slow statement is emitted at WARN level with its duration and (truncated) SQL, and increments the Slow_queries counter:

WARN elyra_server::observ: slow query duration_ms=1636 sql=SELECT COUNT(*) FROM t a, t b

--slow-query-ms 0 (the default) disables slow-query logging. Logs go through the standard tracing subscriber; set RUST_LOG to adjust verbosity.

Prometheus metrics endpoint

Start the server with --metrics-listen to expose the same counters over HTTP in the Prometheus text exposition format:

elyrasql serve --data elyra.edb --metrics-listen 0.0.0.0:9464
# or ELYRASQL_METRICS_LISTEN=0.0.0.0:9464

GET /metrics returns metrics such as:

elyrasql_uptime_seconds 3600
elyrasql_connections_current 2
elyrasql_connections_total 57
elyrasql_questions_total 12045
elyrasql_commands_total{command="select"} 9001
elyrasql_commands_total{command="insert"} 2003
elyrasql_errors_total 4
elyrasql_slow_queries_total 12

GET /health returns 200 OK for liveness probes. Point a Prometheus scrape job at the endpoint:

scrape_configs:
  - job_name: elyrasql
    static_configs:
      - targets: ['elyra-host:9464']

Not yet available

There is no performance_schema; per-query histograms and connection-level metrics beyond the process list are not exposed yet.