Class QuartzSchedulerOptions

Namespace

Quartz

Assembly

Quartz.dll

Strongly typed configuration for the scheduler itself.

public sealed class QuartzSchedulerOptions

Inheritance

object

QuartzSchedulerOptions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Binds from the Scheduler section of the Quartz configuration, and is the typed replacement for the quartz.scheduler.* property keys.

Fields

DefaultInstanceId

The value for InstanceId that marks a scheduler as not participating in a cluster.

public const string DefaultInstanceId = "NON_CLUSTERED"

Field Value

string

DefaultInstanceName

The default value for InstanceName.

public const string DefaultInstanceName = "QuartzScheduler"

Field Value

string

Properties

BatchTriggerAcquisitionFireAheadTimeWindow

How far past the current time a trigger may fire in order to be included in the current acquisition batch.

public TimeSpan BatchTriggerAcquisitionFireAheadTimeWindow { get; set; }

Property Value

TimeSpan

Remarks

The other half of MaxBatchSize; neither batches anything on its own. At the default of Zero a batch holds the triggers due at the same instant and nothing else. Widening it fires triggers early by up to this much, which is what the batching costs.

Context

Values seeded into SchedulerContext when the scheduler is created.

public Dictionary<string, string> Context { get; }

Property Value

Dictionary<string, string>

Remarks

Replaces the quartz.context.key.* property keys.

GenerateInstanceId

When true, InstanceId is generated at startup by the registered IInstanceIdGenerator instead of being taken from InstanceId.

public bool GenerateInstanceId { get; set; }

Property Value

bool

IdleWaitTime

How long the scheduler waits before re-querying the job store when it finds no triggers to fire.

public TimeSpan IdleWaitTime { get; set; }

Property Value

TimeSpan

Remarks

Must be at least one second. Lower values increase job store load for little benefit.

InstanceId

The id of the scheduler, which must be unique within a cluster.

public string InstanceId { get; set; }

Property Value

string

Remarks

Leave at the default to opt out of clustering. Set GenerateInstanceId instead of assigning a literal id when the id should be derived at startup.

InstanceName

The name of the scheduler, which must be unique within the process.

public string InstanceName { get; set; }

Property Value

string

MaxBatchSize

The maximum number of triggers the scheduler acquires in a single batch.

public int MaxBatchSize { get; set; }

Property Value

int

Remarks

Defaults to 1, which is also the point at which a database-backed store needs no cluster-wide lock to acquire: raising it makes every acquisition cycle take the TRIGGER_ACCESS row lock, including cycles that acquire nothing. It pays off where triggers genuinely arrive in bunches, and costs lock traffic where they do not.

This is only the upper bound. What decides the size of a batch is BatchTriggerAcquisitionFireAheadTimeWindow: after the first trigger, only triggers due within that window of it join the batch, and the window defaults to Zero. Raising this alone leaves the effective batch at one trigger for any schedule whose fire times are spread out — the two are one setting in two halves, so move them together.

Must not exceed MaxConcurrency: triggers acquired beyond the number of threads there are to run them on are held by this node, unfireable by any other, until the pool drains.

PropagateTraceContext

Whether the scheduler records the trace context of the call that scheduled a trigger, so that the firing links back to it. On by default.

public bool PropagateTraceContext { get; set; }

Property Value

bool

Remarks

A scheduled job runs minutes, hours or days after the call that asked for it, quite possibly on another node. With this on, a scheduling call made inside an activity leaves the W3C traceparent on the trigger — under TraceParent and TraceState — and the firing's Quartz.Job.Execute span carries an ActivityLink to it. A link and not a parent: the firing is its own trace root, because a trace spanning the wait would be a trace nothing could display.

Turn it off to keep the two reserved keys out of trigger data entirely — for a store whose rows are read by something that does not expect them, or where the extra two entries per trigger are not worth what they buy. Nothing else changes: the execute span is emitted either way.

ShutdownJobInterruption

When a shutting-down scheduler signals cancellation to the jobs still executing.

public ShutdownJobInterruption ShutdownJobInterruption { get; set; }

Property Value

ShutdownJobInterruption

Remarks

This was two independent booleans, one for a shutdown that waits for jobs and one for a shutdown that does not. They were never independent: together they answered a single four-way question, and the pair could be set to a combination — both false — that spelled the default in two ways.