alter_job()

Community Since 1.2.0

Jobs scheduled using the TimescaleDB automation framework run periodically in a background worker. You can change the schedule of these jobs with the alter_job function. To alter an existing job, refer to it by job_id. The job_id runs a given job, and its current schedule can be found in the timescaledb_information.jobs view, which lists information about every scheduled jobs, as well as in timescaledb_information.job_stats. The job_stats view also gives information about when each job was last run and other useful statistics for deciding what the new schedule should be.

Calculate the next start on failure

When a job run results in a runtime failure, the next start of the job is calculated taking into account both its retry_period and schedule_interval. The next_start time is calculated using the following formula:

next_start = finish_time + consecutive_failures * retry_period ± jitter

where jitter (± 13%) is added to avoid the “thundering herds” effect.

To ensure that the next_start time is not put off indefinitely or produce timestamps so large they end up out of range, it is capped at 5*schedule_interval. Also, more than 20 consecutive failures are not considered, so if the number of consecutive failures is higher, then it multiplies by 20.

Additionally, for jobs with fixed schedules, the system ensures that if the next start ( calculated as specified), surpasses the next scheduled execution, the job is executed again at the next scheduled slot and not after that. This ensures that the job does not miss scheduled executions.

There is a distinction between runtime failures that do not cause the job to crash and job crashes. In the event of a job crash, the next start calculation follows the same formula, but it is always at least 5 minutes after the job‘s last finish, to give an operator enough time to disable it before another crash.

Samples

• Reschedule job ID 1000 so that it runs every two days:

  SELECT alter_job(1000, schedule_interval => INTERVAL '2 days');

• Disable scheduling of the compression policy on the conditions hypertable:

  SELECT alter_job(job_id, scheduled => false)
  FROM timescaledb_information.jobs
  WHERE proc_name = 'policy_compression' AND hypertable_name = 'conditions'

• Reschedule continuous aggregate job ID 1000 so that it next runs at 9:00:00 on 15 March, 2020:

  SELECT alter_job(1000, next_start => '2020-03-15 09:00:00.0+00');

• Alter a columnstore policy:

  You can pause and restart a columnstore policy, change how often the policy runs and the job scheduling. To do this:

  -- Find the {JOB} ID for the {COLUMNSTORE} policy:
        SELECT job_id, hypertable_name, config
        FROM timescaledb_information.jobs
        WHERE proc_name = 'policy_compression';
  
  -- For example, to change the compression interval after 30 days instead of 7:
        SELECT alter_job(1000, config => '{"compress_after": "30 days"}');

  However, to change the after or created_before, the compression settings, or the hypertable the policy is acting on, you must remove the columnstore policy and add a new one.

Arguments

The syntax is:

SELECT alter_job(
    job_id = <job_id>,
    schedule_interval = <interval>,
    max_runtime = <interval>,
    max_retries = <integer>,
    retry_period = <interval>,
    scheduled = true | false,
    config = '<jsonb_config>',
    next_start = <timestamptz>,
    if_exists = true | false,
    check_config = '<procedure_name>',
    fixed_schedule = true | false,
    initial_start = <timestamptz>,
    timezone = '<timezone>'
);

Name	Type	Default	Required	Description
job_id	INTEGER	-	✔	The ID of the policy job being modified.
schedule_interval	INTERVAL	24 hours	✖	The interval at which the job runs.
max_runtime	INTERVAL	-	✖	The maximum amount of time the job is allowed to run by the background worker scheduler before it is stopped.
max_retries	INTEGER	-	✖	The number of times the job is retried if it fails.
retry_period	INTERVAL	-	✖	The amount of time the scheduler waits between retries of the job on failure.
scheduled	BOOLEAN	true	✖	Set to false to exclude this job from being run as a background job.
config	JSONB	-	✖	Job-specific configuration, passed to the function when it runs. This includes: verbose_log: boolean, defaults to false. Enable verbose logging output when running the columnstore policy. maxchunks_to_compress: integer, defaults to 0 (no limit). The maximum number of chunks to add to the columnstore during a policy run. recompress: boolean, defaults to true. Recompress partially compressed chunks. compress_after: see add_columnstore_policy. compress_created_before: see add_columnstore_policy.
next_start	TIMESTAMPTZ	-	✖	The next time at which to run the job. The job can be paused by setting this value to infinity, and restarted with a value of now().
if_exists	BOOLEAN	false	✖	Set to true to issue a notice instead of an error if the job does not exist.
check_config	REGPROC	-	✖	A function that takes a single argument, the JSONB config structure. The function is expected to raise an error if the configuration is not valid, and return nothing otherwise. Can be used to validate the configuration when updating a job. Only functions, not procedures, are allowed as values for check_config.
fixed_schedule	BOOLEAN	false	✖	To enable fixed scheduled job runs, set to true.
initial_start	TIMESTAMPTZ	-	✖	Set the time when the fixed_schedule job run starts. For example, 19:10:25-07.
timezone	TEXT	UTC	✖	Address the 1-hour shift in start time when clocks change from Daylight Saving Time to Standard Time. For example, America/Sao_Paulo.
job_name	TEXT	NULL	✖	A name for the job.

When a job begins, the next_start parameter is set to infinity. This prevents the job from attempting to be started again while it is running. When the job completes, whether or not the job is successful, the parameter is automatically updated to the next computed start time.

Note that altering the next_start value is only effective for the next execution of the job in case of fixed schedules. On the next execution, it will automatically return to the schedule.

Returns

Column	Type	Description
job_id	INTEGER	The ID of the job being modified
schedule_interval	INTERVAL	The interval at which the job runs. Defaults to 24 hours
max_runtime	INTERVAL	The maximum amount of time the job is allowed to run by the background worker scheduler before it is stopped
max_retries	INTEGER	The number of times the job is retried if it fails
retry_period	INTERVAL	The amount of time the scheduler waits between retries of the job on failure
scheduled	BOOLEAN	Returns true if the job is executed by the TimescaleDB scheduler
config	JSONB	Job-specific configuration, passed to the function when it runs
next_start	TIMESTAMPTZ	The next time to run the job
check_config	TEXT	The function used to validate updated job configurations
fixed_schedule	BOOLEAN	Whether the job uses a fixed schedule
initial_start	TIMESTAMPTZ	The time when the fixed schedule job run starts
timezone	TEXT	The timezone for the job schedule
application_name	NAME	The application name for the job