Manager commands
Edit this pageSystem · Related pages
sqlstreams manager runRuns the registered installation’s system manager until SIGINT or SIGTERM. It handles eligible maintenance, cursor advancement, scheduled production, metric collection, and alert checks. It does not invoke application consumer handlers.
Syntax
sqlstreams manager run [--metrics-address <address>]| Flag | Default | Description |
|---|---|---|
--metrics-address |
Disabled | Serve a Prometheus /metrics endpoint on this address, such as 127.0.0.1:9464. |
The global connection flags
select the database and installation. Register the system before starting
this command. It has no result document and rejects --output json.
Metrics endpoint
sqlstreams manager run --metrics-address 127.0.0.1:9464The endpoint uses the metrics exporter to read retained measurements on each scrape. It serves plain HTTP at the chosen bind address. A listener failure stops the manager and exits with an error.
Run one scrape target per installation when summing deployment totals. Several manager processes can share the installation, but their exporters read the same retained measurements.
Lifecycle and output
The command writes Info, Warn, and Error logs to stderr. Multiple manager processes coordinate through database worker claims. One holds the manager claim while others wait to claim it. Individual worker targets control how many instances can run.
SIGINT or SIGTERM starts shutdown. A later signal uses the operating
system’s default termination behavior. Shutdown drains the manager, then
the HTTP server, exporter, and pool. A requested stop normally exits 0.
An internal manager run failure logs SQL0065
and retries after backoff while the command remains running. Initial setup
or metrics-server failures exit 1. Invalid usage exits 2.
Related topics
- Run the system manager — Go API and worker lifecycle.
- Worker is not running — Inspect suspension, claims, and failures.
- Export metrics — Collection and freshness behavior.