Summary

SIEM (Security Information and Event Management system) tools don't need a dedicated Semarchy connector or any app-side log export configuration. Two Snowflake-native sources cover xDM Native App activity:

  • Audit views: logins, queries, data access, privilege changes — already supported, read by the SIEM's own Snowflake connector.
  • The account's event table: carries the xDM container's stdout/stderr logs and its platform metrics automatically, once the account has an active event table, with no logExporters setting on the app and no opt-in flag needed.

Confirmed by test on a separate consumer-style account (not the Semarchy builder account), see Test evidence below.


Prerequisite: an active event table

Check the account's event table:

SHOW PARAMETERS LIKE 'EVENT_TABLE' IN ACCOUNT;
  • Value is snowflake.telemetry.events (the default) or another table name → already set, skip to Reading logs.
  • Value is NONE → event collection is off; enable one as ACCOUNTADMIN:
-- Option A (recommended): restore the Snowflake default event table
USE ROLE ACCOUNTADMIN;
ALTER ACCOUNT UNSET EVENT_TABLE;

-- Option B: use a custom event table
USE ROLE ACCOUNTADMIN;
CREATE DATABASE IF NOT EXISTS TELEMETRY_DB;
CREATE SCHEMA IF NOT EXISTS TELEMETRY_DB.LOGS;
CREATE EVENT TABLE IF NOT EXISTS TELEMETRY_DB.LOGS.EVENTS;
ALTER ACCOUNT SET EVENT_TABLE = TELEMETRY_DB.LOGS.EVENTS;

The active event table is account-wide: all telemetry (procedures, UDFs, services) lands there, not just xDM's. Logs are only captured from the moment the event table is active — earlier logs can't be recovered.


Reading xDM container logs

Replace <APP_NAME> with the app's database name (from SHOW APPLICATIONS;):

SELECT resource_attributes:"snow.service.name"::STRING AS service_name,
       resource_attributes:"snow.service.container.name"::STRING AS container_name,
       record_attributes:"log.iostream"::STRING AS stream,
       COUNT(*) AS log_lines,
       MAX(timestamp) AS last_log
FROM SNOWFLAKE.TELEMETRY.EVENTS_VIEW
WHERE timestamp > DATEADD(day, -7, CURRENT_TIMESTAMP())
  AND record_type = 'LOG'
  AND resource_attributes:"snow.database.name" = '<APP_NAME>'
GROUP BY ALL
ORDER BY last_log DESC;

For individual lines (last 24h, optionally filtered to errors):

SELECT timestamp,
       resource_attributes:"snow.service.container.name"::STRING AS container_name,
       value::STRING AS message
FROM SNOWFLAKE.TELEMETRY.EVENTS_VIEW
WHERE timestamp > DATEADD(hour, -24, CURRENT_TIMESTAMP())
  AND record_type = 'LOG'
  AND resource_attributes:"snow.database.name" = '<APP_NAME>'
  -- AND value::STRING ILIKE '%ERROR%'
ORDER BY timestamp DESC
LIMIT 100;


No setting on the app side is needed: both the xdm and repo service containers' stdout/stderr showed up this way in testing. Log lines carry no severity_text, filter on message text for errors. Logs can also be browsed in Snowsight under Monitoring » Traces & Logs, filtered to the app's database.


Reading xDM platform metrics

The same event table also carries metric records (record_type = 'METRIC'):

SELECT record_attributes:"metric.name"::STRING AS metric_name, COUNT(*) AS n, MAX(timestamp) AS last_ts
FROM SNOWFLAKE.TELEMETRY.EVENTS_VIEW
WHERE timestamp > DATEADD(hour, -1, CURRENT_TIMESTAMP())
  AND record_type = 'METRIC'
  AND resource_attributes:"snow.database.name" = '<APP_NAME>'
GROUP BY 1 ORDER BY 1;

The documented path to turn these on is the app's own reload procedure:

CALL <APP_NAME>.XDM_PUBLIC.reload_app({'platform_metrics': true}::VARIANT);



Who can read it

Snowflake's documented requirement for the default event table:

Event table

Required privilege

Default (SNOWFLAKE.TELEMETRY.EVENTS)

Application role SNOWFLAKE.EVENTS_VIEWER (read-only) or SNOWFLAKE.EVENTS_ADMIN (broader); ACCOUNTADMIN also has access

Custom event table

SELECT on the table, plus USAGE on its database and schema