Embedded API

From Obsidian Scheduler
Revision as of 23:59, 16 March 2014 by Carfey (talk | contribs) (→Add a Job)
Jump to navigationJump to search

Obsidian 2.3 introduced a new unified Embedded API which contains all the same actions and semantics as the REST API. It replaces the old Legacy API.

This API is accessible through Java, and can be used in software environments where the REST API is unavailable or undesired.

Javadoc is also available to supplement this page.


Overview

The Obsidian Embedded API allows your application to access Obsidian data, schedule jobs, and control Obsidian in a variety of ways.

The API gives users the power to do things like initialize scheduled jobs at deployment time, trigger jobs based on events in your applications, and expose pieces of the Obsidian API in their own custom user interfaces.

Accessing the API

The Embedded API is written in Java and can be be accessed within the context of a running Obsidian node, although the scheduler itself need not be active. You can even call API operations from within jobs, which you can use to get custom chaining or workflow behaviour, or update configuration on certain types of events detected in a job.

In addition, it can be embedded in applications that are not running Obsidian, simply by including Obsidian's dependent libraries and properties file.

Authorization

The Embedded API is not secured, though all actions accept an audit user which is linked to actions which modify Obsidian state. Any code that has access to the Obsidian database via its defined credentials can manipulate Obsidian job configurations.

Transactions

By default, when you invoke actions within the API, each individual call is in its own transaction which is committed when successful, and rolls back on failure.

However, you can wrap multiple API calls into a single transaction, so you get one unit of work which either is fully committed or rolled back. An example is shown below. For full semantics, see AbstractAPIManager.withTransaction. 

final String auditUser = "Bob";

List<HostDetail> updated = new HostManager().withTransaction(auditUser, new Callable<<HostDetail>>() {
    
    public List<HostDetail> call() throws Exception {
        
        // disable some hosts in an atomic manner (does not have to be the same instance of even type of Manager class)
        HostDetail hostA = new HostManager().updateHost("hostA", new HostUpdateRequest().withEnabled(false), auditUser );
        HostDetail hostB = new HostManager().updateHost("hostB", new HostUpdateRequest().withEnabled(false), auditUser );
            
        // this is within the same transaction
        new JobManager().resubmitRuntime(jobToResubmit, auditUser);

        return Arrays.asList(hostA, hostB);
    }
});
     
System.out.println("Updated hosts: " + updated);

API Basics

The API is exposed through Manager classes in the package com.carfey.ops.api.embedded:

  • JobManager - Create and manage jobs, job conflicts and chaining.
  • RuntimeManager - List job runtime history, resubmit jobs, preview scheduled times, etc.
  • HostManager - Manage and list active scheduling hosts (i.e. nodes).
  • CustomCalendarManager - Manage custom calendars.


To invoke an API method, simply create a new Manager instance, and invoke the appropriate method: 

RuntimeListing recentRuntimes = new RuntimeManager().listRuntimes(new RuntimeListingParameters());

You can reuse manager instances if you like, but it is ultimately up to your preference.

Exceptions

API calls generally throw three types of exceptions:

  • ValidationException - some sort of validation error occurred.
  • MissingEntityException - The requested entity could not be found by the ID supplied.
  • All other Exception types - some other type of error that you probably can't do much with (database issues, etc.).

When integrating the API into your application, you may wish to apply appropriate handling to ValidationException and MissingEntityException, but otherwise you can treat exceptions as a generic server errors.

Enumerations

Below are valid values for commonly used fields in the API. These are also used by the REST API.

Corresponding Java enums can be found in the package com.carfey.ops.api.enums.

Note on compatibility: Prior to Obsidian 2.5, some request classes used String values rather than Java enums. This is no longer the case.

Note on naming: Prior to Obsidian 2.5, input values for REST endpoints could use spaces in place of underscores for enumeration values. This is no longer the case.

Job Status

  • ENABLED
  • DISABLED
  • UNSCHEDULED_ACTIVE
  • CHAIN_ACTIVE
  • AD_HOC_ACTIVE


Job History Status

  • READY
  • RUNNING
  • COMPLETED
  • FAILED
  • MISSED
  • DIED
  • CONFLICTED
  • OVERLAPPED
  • ABANDONED
  • CONFLICT_MISSED
  • CHAIN_SKIPPED


Job Recovery Type

  • NONE
  • LAST
  • ALL
  • CONFLICTED


Job Parameter Type

  • STRING
  • INTEGER
  • LONG
  • DECIMAL
  • BOOLEAN
  • CLASS


Job Chain Status

  • ABANDONED
  • COMPLETED
  • CONDITIONAL
  • DIED
  • FAILED
  • MISSED
  • OVERLAPPED
  • CHAIN_SKIPPED


Job Chain Condition Operator

  • EQUALS
  • NOT_EQUALS
  • IN
  • NOT_IN
  • EXISTS
  • NOT_EXISTS
  • REGEXP
  • STARTS_WITH
  • ENDS_WITH
  • CONTAINS
  • GREATER_THAN
  • LESS_THAN
  • GREATER_THAN_OR_EQUAL
  • LESS_THAN_OR_EQUAL

JobManager API

JobManager is used to manage job configurations, conflicts and chaining. If you are looking for ways to view or manage job runtimes, see RuntimeManager.

List Jobs

You can list or search existing jobs using JobManager.listJobs(), which accepts an optional JobListingParameters and returns a JobListing.

You can limit the returned jobs by providing a JobListingParameters.


JobListingParameters Fields

Field Required? Notes
activeStatuses N Restricts the preview to the selected a valid JobStatus values
effectiveDate N If querying by activeStatuses, this allows you to indicate what point in time to compare against the job status. Defaults to next minute.
hosts N If specified, only jobs that run on the specified host name(s) are included.
nicknames N If specified, only jobs matching the supplied nicknames are returned. Wildcards may be included to support partial matches by using %, or exact literals can be used. For example, to find all jobs containing the word "order", use "%order%".
filterParameters N If specified, values supplied in this map can be used to match on job parameter values, either custom or defined. If multiple values for the same name (i.e. key) are supplied, a job is matched if any of its configured values match one of the supplied values. If multiple names are used, each must have a matching value for the job to be returned.

Usage note: This can be used to tag jobs with searchable metadata by configuring custom parameters. For example, if jobs belong to logical groups, you may create a custom parameter on applicable jobs named "group" and use filterParameters to locate jobs belonging to specific groups, such as "customer" or "order".


REST equivalent: GET a list of jobs

Get a Job's Details

To get full job information, including all historical schedules and parameter information, call JobManager.getJob() with the appropriate job ID. The method will return a JobDetail, or throw a MissingEntityException if it does not exist.

REST equivalent: GET details of an existing job

Add a Job

You can add a new job using JobManager.addJob(), which accepts a JobCreationRequest. The method will return the job as a JobDetail.

When you create the job, you provide an initial schedule.

Note: When supplying effective and end dates for schedules, seconds must be omitted. To do so, use com.carfey.jdk.lang.DateTime.clearSeconds(), which returns a copy of the date with seconds removed.


JobCreationRequest Fields

Field Required? Notes
jobClass Y Fully qualified class name of the job. Max 255 chars.
nickname Y Unique nickname for the job. Max 255 chars.
pickupBufferMinutes N Pickup buffer minutes which defaults to 2. Integer greater than zero.
recoveryType Y The JobRecoveryType to use for the job.
state Y Initial schedule's JobStatus.
schedule Y/N If state is ENABLED, the mandatory cron-style schedule for the job. If not ENABLED, this should be omitted.
effectiveDate N Optional effective date for the initial schedule, with no seconds specified. If not set, this defaults to next minute. Until this date is reached, the job is DISABLED.
endDate N Optional end date for the initial schedule, with no seconds specified. If set, the job will become DISABLED after this date passes.
customCalendarId N Optional custom calendar id.
hosts N Zero or more host names that this job may run on. If none set, the job may run on any host.
minExecutionDuration N The minimum expected job runtime. Format is an integer greater than zero immediately followed by "s", "m" or "h". Example: "15m".
maxExecutionDuration N The maximum expected job runtime. Format is an integer greater than zero immediately followed by "s", "m" or "h". Example: "15m".
autoRetryCount N Number of auto retries on non-interrupted execution failure. Defaults to 0, which means none are desired.
chainAll N Boolean indicating whether all chained instances are triggered when job is currently running. Otherwise, only one newly chained record is created. Defaults to false.
parameters Y/N Zero or more parameter definitions. If a job defines required parameters with the @Configuration annotation, a job will fail to create unless they are supplied. Otherwise, this field is optional. Parameter definitions must have values for name, type and value, where type is a valid ParameterType. To define multiple values for a single parameter name, simply include multiple items in the parameters collection. As of 2.5, values may contain global parameter references (e.g. Template:GlobalParameterName.


REST equivalent: POST a new job

Update a Job

You can update an existing job using JobManager.updateJob(), which accepts a job ID and a JobUpdateRequest. The method will return the new state of the job as a JobDetail, or throw a MissingEntityException if it does not exist.

This does not support schedule changes or additions. For schedule changes, see Add a New Schedule to a Job.

This method will only update fields that are supplied in the JobUpdateRequest. You may update one or more fields as desired.

JobUpdateRequest Format

Field Required? Notes
jobClass N Fully qualified class name of the job. Max 255 chars.
nickname N Unique nickname for the job. Max 50 chars.
pickupBufferMinutes N Pickup buffer minutes. Integer greater than zero.
recoveryType N The JobRecoveryType to use for the job.
hosts N Zero or more host names that this job may run on. If none set, the job may run on any host.
minExecutionDuration N The minimum expected job runtime. Format is an integer greater than zero immediately followed by "s", "m" or "h". Example: "15m".
maxExecutionDuration N The maximum expected job runtime. Format is an integer greater than zero immediately followed by "s", "m" or "h". Example: "2h".
autoRetryCount N Number of auto retries on non-interrupted execution failure. 0 if none are desired.
chainAll N Boolean indicating whether all chained instances are triggered when job is currently running. Otherwise, only one newly chained record is created.
parameters N Zero or more parameter definitions. If a job defines required parameters with the @Configuration annotation, a job will fail to create unless they are supplied. Otherwise, this field is optional. Note that a null value will result in existing parameters being preserved, while an empty list is considered replacing existing parameters. Parameter definitions must have values for name, type and value, where type is a valid ParameterType. To define multiple values for a single parameter name, simply include multiple items in the parameters collection.


REST equivalent: PUT updates to an existing job

Delete a Job

You can delete an existing job using JobManager.deleteJob(), which accepts a job ID and cascade flag. The method will return the final state of the job before deletion as a JobDetail, or throw a MissingEntityException if it does not exist.


Parameters

Field Required? Notes
cascade Y If set to true, all job conflict and chain definitions for this job will also be deleted. If set to false, any existing job conflicts or chain definitions will cause the request to fail.


REST equivalent: DELETE an existing job

List a Job's Schedules

You can access the full history of a job's schedules via JobManager.listJobSchedules(), which will return JobScheduleListing.

Note that the full schedule history can also be obtained via the schedules field of the JobDetail returned by JobManager.getJob().

REST equivalent: GET a list of an existing job's schedules

Add a New Schedule to a Job

You can add a new schedule to a job via JobManager.addJobSchedule(), which accepts a ScheduleCreationRequest.

This may be used to immediately change a job's scheduling state, or to schedule a future change. Creating a new schedule automatically splits and merges existing schedules. For example, if you have an enabled job and you disabled it for a day, the job will automatically re-enable after that day.

Note: When supplying effective and end dates for schedules, seconds must be omitted. To do so, use com.carfey.jdk.lang.DateTime.clearSeconds(), which returns a copy of the date with seconds removed.


ScheduleCreationRequest Fields

Field Required? Notes
state Y The schedule's JobStatus.
schedule Y/N If state is ENABLED, the mandatory cron-style schedule for the job. If not ENABLED, this should be omitted.
effectiveDate N Optional effective date for the schedule, with no seconds specified. If not set, this defaults to next minute. Until this date is reached, the job is DISABLED.
endDate N Optional end date for the schedule, with no seconds specified. If set, the job will become DISABLED after this date passes.
customCalendarId N Optional custom calendar for schedule.


REST equivalent: POST a new schedule to an existing job

List Job Chains

You can list or search existing job chains using JobManager.listChains(), which accepts an optional JobChainListingParameters and returns a JobChainListing.

You can limit the returned job chains by providing a JobChainListingParameters instance with the fields below.


JobChainListingParameters Fields

Field Required? Notes
active N Limits results to those matching the active flag (true/false).
sourceJobId N Limits results to those matching the supplied source job ID.
targetJobId N Limits results to those matching the supplied target job ID.


REST equivalent: GET a list of job chains

Add a Job Chain

You can add a new job chain using JobManager.addChain(), which accepts a JobChainUpdateRequest. The method will return a JobChain.


JobChainUpdateRequest Fields

Field Required? Notes
sourceJobId Y ID of the source job.
targetJobId Y ID of the target job to chain
schedule N Optional schedule that constrains when the job chain triggers.
triggerStates Y One or more JobChainStatus. Note that CONDITIONAL and COMPLETED types cannot be used on the same chain.
resultConditions Y/N Conditions based on job results that apply to the CONDITIONAL trigger state. Must be supplied only when that state is used, in which case at least one condition must be supplied. Result conditions consist of a variableName, an operator as a JobChainConditionOperator, and for most operators, a list of values. The EXISTS and NOT EXISTS operators do not use values, so they must not be supplied. Operators IN and NOT IN support one or more values, and all other operators accept a single value in the values list.


REST equivalent: POST a new job chain

Update a Job Chain

You can add update an existing job chain using JobManager.updateChain(), which accepts a JobChainUpdateRequest. The method will return the updated JobChain.

The format of JobChainUpdateRequest is identical to that of Add a Job Chain.

REST equivalent: PUT updates to an existing job chain

List Conflicts

You can access all configured job conflict configuration via JobManager.listConflicts(), which will return ConflictListing.

The return value includes all job conflict sets, with each set in order of priority. Non-conflicted jobs are also returned.

REST equivalent: GET a list of job conflicts

Update Conflicts

To update the configured job conflict sets, use JobManager.updateConflicts() with a ConflictUpdateRequest which will return the updated ConflictListing.

The request replaces the current job conflict configuration with the supplied configuration. The return value includes all job conflict sets, with each set in order of priority. Non-conflicted jobs are also returned.


ConflictUpdateRequest Format

Field Required? Notes
conflicts N A list of ordered sets containing job IDs. Each inner set contains jobs that conflict with each other, in order of execution precedence. Jobs that do not conflict with any other jobs are simply omitted from this list. A job can only exist in one conflict set. To remove all job conflicts, an empty list can be supplied for this field.

REST equivalent: PUT updates to job conflicts

List a Specific Job's Conflicts

You can return conflicts for a specific job with JobManager.listJobConflicts(long), which will return JobConflictListing.

Conflicting jobs are returned in priority order, including the job for which this request was made, in order that its priority within the set can be determined. Note that if the job has no conflicts, the returned conflicting jobs list will be empty, and will not contain the requested job.

REST equivalent: GET a list of conflicts for a specific job


Update Global Parameters

To update the configured global parameters, use JobManager.updateGlobalParameters() with a GlobalParameterUpdateRequest which will return the updated GlobalParameterListing.

The request replaces the current global parameters with the supplied configuration. Supplying an empty list of parameters will result in all global parameters being deleted.

Note: Calls to remove or alter global parameters may fail if jobs that use them do not pass parameter validation as a result of the change. This can be caused by removing a referenced global parameter or values that cannot be interpreted as the appropriate type in a job.


GlobalParameterListing Format

Field Required? Notes
parameters N Zero or more GlobalParameter instances. Each parameter must have values for name, type and values, where type is a valid ParameterType.

REST equivalent: PUT updates to global parameters

List Global Parameters

You can list all configured global parameters using listGlobalParameters(), which will return a GlobalParameterListing.


REST equivalent: GET a list of configured global parameters

RuntimeManager API

RuntimeManager is used to manage and view scheduled job runtimes (i.e. history). If you are looking for ways to view or manage jobs, see JobManager.

List Scheduled Runtimes

To get a list of scheduled or completed job runtimes (i.e. history), use RuntimeManager.listRuntimes(), which accepts an optional RuntimeListingParameters instance. The results in the returned RuntimeListing are ordered roughly according to when they were created, but ordering is not guaranteed to be in order of scheduled time.

Note: The nextPageStartKey field in the RuntimeListing indicates that there were too many results to return (i.e. exceeded maxRecords as configured in the Admin System tab). To fetch the next page of results, invoke the same method with the startKey field on RuntimeListingParameters set to the returned nextPageStartKey.


RuntimeListingParameters Fields

Field Required? Notes
jobIds N Restricts the search to the selected jobs.
statuses N Restricts the search to the selected JobRuntimeStatus values.
hosts N If specified, only job runtimes that are assigned to the specified host(s) are included.
startDate N Start date for the job runtimes to return (inclusive). Defaults to 24 hours ago.
endDate N End date for the job runtimes to return (inclusive). Defaults to a day after the start time. Must be after the start time.
startKey N If requesting the next page of results from a previous call, set it to the returned nextPageStartKey.


REST equivalent: GET a list of scheduled runtimes

Get Details on a Scheduled or Completed Runtime

To get full details on a scheduled or completed job runtime, including job results and one-time run configuration, use RuntimeManager.getRuntime(). The method returns a RuntimeResult, which has a nested RuntimeDetail containing the full job output and one-time run parameters, or throws a MissingEntityException if it does not exist.


REST equivalent: GET details of an existing scheduled or completed runtime

Preview Runtimes

To preview future runtimes for one or more jobs, use RuntimeManager.listRuntimePreview(), which accepts an optional RuntimePreviewParameters instance to filter the results. The method returns a RuntimePreviewListing.

This method is useful to see when jobs will run during a given time period. Note that these are an estimate of runtimes and cannot account for overlapped jobs, schedule changes or other issues that may result in altered execution times. Results are ordered by scheduled time descending.

Note: The capped field in the returned RuntimePreviewListing indicates that there were too many results to return (i.e. exceeded maxRecords as configured in the Admin System tab). If you are hitting this condition, try limiting your date range or other parameters. Due to the nature of the runtime preview, paging is not feasible.


RuntimePreviewParameters Fields

Field Required? Notes
jobIds N Restricts the preview to the selected jobs.
startDate N Start date for the runtimes to preview (inclusive). Defaults to the current minute.
endDate N End date for the runtimes to preview (inclusive). Defaults to a day after the start time. Must be after the start time.


REST equivalent: GET a list of runtime previews

Schedule a New Runtime for a Job

To submit an ad hoc job run (executed immediately), or a one-time run scheduled for a later time, use RuntimeManager.submitRuntime(). A RuntimeSubmissionRequest must be supplied. The method returns a RuntimeSubmissionResult if it succeeds, or throws a MissingEntityException if the specified job ID does not exist.

Note: The job must be in a valid state to allow for execution (i.e. UNSCHEDULED_ACTIVE, ENABLED, or AD_HOC_ACTIVE).


RuntimeSubmissionRequest Fields

Field Required? Notes
scheduledTime N The scheduled time, when the request is for a scheduled one-time run. If not supplied, the runtime is submitted for immediate execution as an ad hoc job.
parameters N Zero or more parameter definitions. Parameter definitions must have values for name, type and value, where type is a valid ParameterType. To define multiple values for a single parameter name, simply include multiple items in the parameters collection. If the parameter name matches a parameter defined for the job, it must be of the same type.


REST equivalent: POST a new scheduled runtime for an existing job

Resubmit a Failed Job Runtime

If a job execution fails and you wish to resubmit it, use RuntimeManager.resubmitRuntime() with the appropriate job runtime ID. The method returns a RuntimeResubmissionResult if the action succeeds, or throws a MissingEntityException if the specified job runtime ID does not exist.

REST equivalent: POST a resubmission request for a failed job runtime

Interrupt a Running Job Runtime

To interrupt a currently running job runtime, use RuntimeManager.interruptRuntime() with the appropriate job runtime ID. A JobInterruptResult is returned upon success, or a MissingEntityException is thrown if the specified job runtime ID does not exist.

This request can only be made once successfully. An interruption request will result in the job being terminated, as long as it does not terminate naturally very soon after the request is made, and the job is capable of shutting down. Not all jobs can be terminated. See Interruptable Jobs for full details.

REST equivalent: POST an interruption request to kill a running job

HostManager API

HostManager is used to manage and view scheduling hosts.

List Scheduling Hosts

To get a list of known hosts which are running or recently shut down abnormally, use HostManager.listHosts(), which returns a HostListing.

REST equivalent: GET a list of known scheduling hosts

Get a Specific Host

To get details on a host by ID or name, use HostManager.getHost(id) or or HostManager.getHost(String).

The method returns a HostDetail, or throws a MissingEntityException if the specified host ID or name does not exist.

Rest equivalent: GET details on an existing scheduling host

Update a Host

To enable or disable a scheduling host, use HostManager.updateHost(long) or HostManager.updateHost(String) with a HostUpdateRequest. Both methods return a HostDetail, or throw a MissingEntityException if the specified host ID or name does not exist.

HostUpdateRequest Fields

Field Required? Notes
enabled Y Should this host should be enabled?


Rest equivalent: PUT updates to an existing scheduling host

CustomCalendarManager API

The CustomCalendarManager allows for managing of Custom Calendars.

List Calendars

To list existing custom calendars, use CustomCalendarManager.listCalendars(), which returns a CustomCalendarListing.

Rest equivalent: GET a list of custom calendars

Get Details on a Calendar

To get details on an existing custom calendar, use CustomCalendarManager.getCalendar(), which returns a CustomCalendar, or throws a MissingEntityException if the specified calendar ID does not exist.

Rest equivalent: GET details on an existing custom calendar

Add a Calendar

To add a new custom calendar, use CustomCalendarManager.addCalendar() with a CustomCalendarUpdateRequest. The method returns a CustomCalendar.


CustomCalendarUpdateRequest Format

Field Required? Notes
name Y Calendar name
dates Y Dates to exclude


Rest equivalent: POST a new custom calendar

Update a Calendar

To update an existing calendar, use CustomCalendarManager.updateCalendar() with a CustomCalendarUpdateRequest. The method returns a CustomCalendar or throws a MissingEntityException if the specified calendar ID does not exist.

The format of CustomCalendarUpdateRequest is identical to that of Add a Calendar.

Rest equivalent: PUT updates to an existing custom calendar

Retrieved from "https://wiki.obsidianscheduler.com/doc/index.php?title=Embedded_API&oldid=1573"