Skip to main content

​
mdblog()

Returns recent log messages from the Metadb system. You can optionally specify a time interval to filter logs.

​
Signature

​
interval
interval
The time period to retrieve logs from (e.g., ‘1 hour’, ‘30 minutes’). If omitted, returns logs from the past 24 hours.
​
Return Type
table
Returns a table with the following columns:
​
log_time
timestamptz(3)
The timestamp when the log entry was created (with millisecond precision)
​
error_severity
text
The severity level of the log message: INFO, WARNING, ERROR, or FATAL (ordered from least to most severe)
​
message
text
The log message content

​
Usage

-- Logs from past hour
select * from mdblog('1 hour');

-- Logs from past 24 hours (default)
select * from mdblog();

-- Logs from past 30 minutes
select * from mdblog('30 minutes');

​
Example Output

          log_time          | error_severity |              message              
----------------------------+----------------+-----------------------------------
 2024-03-11 10:15:23.456-05 | INFO           | Stream processor started
 2024-03-11 10:20:45.123-05 | WARNING        | High memory usage detected
 2024-03-11 10:25:10.789-05 | ERROR          | Connection timeout to Kafka broker
(3 rows)

​
Use Cases

Check recent errors when investigating system problems:
select * from mdblog('1 hour')
where error_severity in ('ERROR', 'FATAL')
order by log_time desc;
Monitor warning messages to catch potential issues early:
select log_time, message
from mdblog('24 hours')
where error_severity = 'WARNING';
Search logs for specific events or patterns:
select * from mdblog('1 hour')
where message like '%snapshot%'
order by log_time;
Log entries are also stored in the metadb.log table. Use mdblog() for quick access to recent entries, or query metadb.log directly for more advanced filtering.

​
ps()

Returns information about currently running query processes, similar to the Unix ps command.

​
Signature

​
Return Type
table
Returns a table with the following columns:
​
dbname
text
The database name the query is running against
​
username
text
The user executing the query
​
state
text
The current state of the process (e.g., ‘active’, ‘idle’, ‘idle in transaction’)
​
realtime
text
The elapsed wall-clock time the query has been running
​
query
text
The SQL query text being executed

​
Usage

select * from ps();

​
Example Output

  dbname  | username | state  | realtime |              query               
----------+----------+--------+----------+----------------------------------
 metadb   | admin    | active | 00:01:23 | select * from library.patron__
 metadb   | analyst  | idle   | 00:00:00 | 
(2 rows)

​
Use Cases

See which queries are currently executing:
select username, realtime, query
from ps()
where state = 'active';
Identify queries that have been running for an extended period:
select * from ps()
where state = 'active'
order by realtime desc;
See what each user is currently doing:
select username, count(*) as connections, 
       count(*) filter (where state = 'active') as active_queries
from ps()
group by username;
The ps() function shows processes across all databases. Filter by dbname if you only want to see activity in a specific database.

​
Severity Levels

Log messages in mdblog() use the following severity levels (from least to most severe):
​
INFO
Informational messages about normal operations (e.g., “Stream processor started”)
​
WARNING
Warnings about potential issues that don’t prevent operation (e.g., “High memory usage”)
​
ERROR
Errors that affect specific operations but don’t crash the system (e.g., “Connection timeout”)
​
FATAL
Critical errors that may cause system shutdown or data loss

metadb.log Table

Query the complete log history

System Info Functions

Get version and user information

metadb.table_update

Track table update statistics

Build docs developers (and LLMs) love

Get started for freeTalk to us