This document is for an in-development version of Galaxy. You can alternatively view this page in the latest release if it exists or view the top of the latest release's documentation.

galaxy.jobs package

Support for running a tool in Galaxy via an internal job management system

class galaxy.jobs.JobDestination(**kwds)[source]

Bases: galaxy.util.bunch.Bunch

Provides details about where a job runs

class galaxy.jobs.JobToolConfiguration(**kwds)[source]

Bases: galaxy.util.bunch.Bunch

Provides details on what handler and destination a tool should use

A JobToolConfiguration will have the required attribute ‘id’ and optional attributes ‘handler’, ‘destination’, and ‘params’

galaxy.jobs.config_exception(e, file)[source]
class galaxy.jobs.JobConfiguration(app)[source]

Bases: galaxy.util.handlers.ConfiguresHandlers

A parser and interface to advanced job management features.

These features are configured in the job configuration, by default, job_conf.xml

DEFAULT_BASE_HANDLER_POOLS = ('job-handlers',)
JOB_RESOURCE_CONDITIONAL_XML = '<conditional name="__job_resource">\n <param name="__job_resource__select" type="select" label="Job Resource Parameters">\n <option value="no">Use default job resource parameters</option>\n <option value="yes">Specify job resource parameters</option>\n </param>\n <when value="no"/>\n <when value="yes"/>\n </conditional>'

Parse the job configuration XML.

get_tool_resource_xml(tool_id, tool_type)[source]

Given a tool id, return XML elements describing parameters to insert into job resources.

Tool id:A tool ID (a string)
Tool type:A tool type (a string)
Returns:List of parameter elements.

The default JobToolConfiguration, used if a tool does not have an explicit defintion in the configuration. It consists of a reference to the default handler and default destination.

Returns:JobToolConfiguration – a representation of a <tool> element that uses the default handler and destination

Get all configured JobToolConfigurations for a tool ID, or, if given a list of IDs, the JobToolConfigurations for the first id in ids matching a tool definition.


You should not mix tool shed tool IDs, versionless tool shed IDs, and tool config tool IDs that refer to the same tool.

Parameters:ids (list or str.) – Tool ID or IDs to fetch the JobToolConfiguration of.
Returns:list – JobToolConfiguration Bunches representing <tool> elements matching the specified ID(s).

Example tool ID strings include:

  • Full tool shed id: toolshed.example.org/repos/nate/filter_tool_repo/filter_tool/1.0.0
  • Tool shed id less version: toolshed.example.org/repos/nate/filter_tool_repo/filter_tool
  • Tool config tool id: filter_tool

Given a destination ID or tag, return the JobDestination matching the provided ID or tag

Parameters:id_or_tag (str) – A destination ID or tag.
Returns:JobDestination – A valid destination

Destinations are deepcopied as they are expected to be passed in to job runners, which will modify them for persisting params set at runtime.


Given a destination ID or tag, return all JobDestinations matching the provided ID or tag

Parameters:id_or_tag (str) – A destination ID or tag.
Returns:list or tuple of JobDestinations

Destinations are not deepcopied, so they should not be passed to anything which might modify them.


Load all configured job runner plugins

Returns:list of job runner plugins

Given a collection of handlers or destinations, indicate whether the collection represents a tag or a real ID

Parameters:collection (tuple or list) – A representation of a destination or handler

Given a collection of handlers or destinations, indicate whether the collection represents a tag or a real ID

Parameters:collection (tuple or list) – A representation of a destination or handler

Converts legacy (from a URL) destinations to contain the appropriate runner params defined in the URL.

Parameters:job_runners (list of job runner plugins) – All loaded job runner plugins.
class galaxy.jobs.HasResourceParameters[source]

Bases: object

class galaxy.jobs.JobWrapper(job, queue, use_persisted_destination=False)[source]

Bases: galaxy.jobs.HasResourceParameters

Wraps a ‘model.Job’ with convenience methods for running processes and state management.

__init__(job, queue, use_persisted_destination=False)[source]

Shell fragment to inject dependencies.


Remove the job after it is complete, should return “always”, “onsuccess”, or “never”.


Provide an extension point to disable this isolation, Pulsar builds its own job script so this is not needed for remote jobs.


Return the JobDestination that this job will use to run. This will either be a configured destination, a randomly selected destination if the configured destination was a tag, or a dynamically generated destination from the dynamic runner.

Calling this method for the first time causes the dynamic runner to do its calculation, if any.


Restore the dictionary of parameters from the database.


Prepare the job to run by creating the working directory and the config files.

fail(message, exception=False, stdout='', stderr='', exit_code=None)[source]

Indicate job failure by setting state and message on all output datasets.

pause(job=None, message=None)[source]
change_state(state, info=False, flush=True, job=None)[source]
set_runner(runner_url, external_id)[source]
set_job_destination(job_destination, external_id=None, flush=True, job=None)[source]

Persist job destination params in the database for recovery.

self.job_destination is not used because a runner may choose to rewrite parts of the destination (e.g. the params).

get_destination_configuration(key, default=None)[source]

Get a destination parameter that can be defaulted back in app.config if it needs to be applied globally.

finish(stdout, stderr, tool_exit_code=None, check_output_detected_state=None, remote_metadata_directory=None)[source]

Called to indicate that the associated command has been run. Updates the output datasets based on stderr and stdout from the command, and the contents of the output files.

check_tool_output(stdout, stderr, tool_exit_code, job)[source]
get_dataset_finish_context(job_context, output_dataset_assoc)[source]
setup_external_metadata(exec_dir=None, tmp_dir=None, dataset_files_path=None, config_root=None, config_file=None, datatypes_config=None, resolve_metadata_dependencies=False, set_extension=True, **kwds)[source]

Destination for outputs marked as from_work_dir. This is the normal case, just copy these files directly to the ulimate destination.

class galaxy.jobs.TaskWrapper(task, queue)[source]

Bases: galaxy.jobs.JobWrapper

Extension of JobWrapper intended for running tasks. Should be refactored into a generalized executable unit wrapper parent, then jobs and tasks.

__init__(task, queue)[source]

Prepare the job to run by creating the working directory and the config files.

fail(message, exception=False)[source]
change_state(state, info=False, flush=True, job=None)[source]
set_runner(runner_url, external_id)[source]
finish(stdout, stderr, tool_exit_code=None, **kwds)[source]

Called to indicate that the associated command has been run. Updates the output datasets based on stderr and stdout from the command, and the contents of the output files.

get_dataset_finish_context(job_context, dataset)[source]
setup_external_metadata(exec_dir=None, tmp_dir=None, dataset_files_path=None, config_root=None, config_file=None, datatypes_config=None, set_extension=True, **kwds)[source]

Destination for outputs marked as from_work_dir. These must be copied with the same basenme as the path for the ultimate output destination. This is required in the task case so they can be merged.

class galaxy.jobs.ComputeEnvironment[source]

Bases: object

Definition of the job as it will be run on the (potentially) remote compute server.


Output unqualified filenames defined by job.


Output DatasetPaths defined by job.


Input DatasetPaths defined by job.


Job working directory (potentially remote)


Directory containing config files (potentially remote)


os.path.sep for the platform this job will execute in.


Absolute path to dump new files for this job on compute server.


Absolute path to tool files for this job on compute server.


Location of the version file for the underlying tool.


Return a function that takes in a value, determines if it is path to be rewritten (will be passed non-path values as well - onus is on this function to determine both if its input is a path and if it should be rewritten.)


Home directory of target job - none if HOME should not be set.


Temp directory of target job - none if HOME should not be set.

class galaxy.jobs.SimpleComputeEnvironment[source]

Bases: object

class galaxy.jobs.SharedComputeEnvironment(job_wrapper, job)[source]

Bases: galaxy.jobs.SimpleComputeEnvironment

Default ComputeEnviornment for job and task wrapper to pass to ToolEvaluator - valid when Galaxy and compute share all the relevant file systems.

__init__(job_wrapper, job)[source]
class galaxy.jobs.NoopQueue[source]

Bases: object

Implements the JobQueue / JobStopQueue interface but does nothing

put(*args, **kwargs)[source]
class galaxy.jobs.ParallelismInfo(tag)[source]

Bases: object

Stores the information (if any) for running multiple instances of the tool in parallel on the same set of inputs.




galaxy.jobs.command_factory module

galaxy.jobs.command_factory.build_command(runner, job_wrapper, container=None, modify_command_for_container=True, include_metadata=False, include_work_dir_outputs=True, create_tool_working_directory=True, remote_command_params={}, metadata_directory=None)[source]

Compose the sequence of commands necessary to execute a job. This will currently include:

  • environment settings corresponding to any requirement tags
  • preparing input files
  • command line taken from job wrapper
  • commands to set metadata (if include_metadata is True)

galaxy.jobs.datasets module

Utility classes allowing Job interface to reason about datasets.

class galaxy.jobs.datasets.DatasetPath(dataset_id, real_path, false_path=None, false_extra_files_path=None, mutable=True)[source]

Bases: object

__init__(dataset_id, real_path, false_path=None, false_extra_files_path=None, mutable=True)[source]
with_path_for_job(false_path, false_extra_files_path=None)[source]

Clone the dataset path but with a new false_path.

class galaxy.jobs.datasets.DatasetPathRewriter[source]

Bases: object

Used by runner to rewrite paths.

rewrite_dataset_path(dataset, dataset_type)[source]

Dataset type is ‘input’ or ‘output’. Return None to indicate not to rewrite this path.

class galaxy.jobs.datasets.NullDatasetPathRewriter[source]

Bases: object

Used by default for jobwrapper, do not rewrite anything.

rewrite_dataset_path(dataset, dataset_type)[source]

Keep path the same.

class galaxy.jobs.datasets.OutputsToWorkingDirectoryPathRewriter(working_directory)[source]

Bases: object

Rewrites all paths to place them in the specified working directory for normal jobs when Galaxy is configured with app.config.outputs_to_working_directory. Job runner base class is responsible for copying these out after job is complete.

rewrite_dataset_path(dataset, dataset_type)[source]

Keep path the same.

class galaxy.jobs.datasets.TaskPathRewriter(working_directory, job_dataset_path_rewriter)[source]

Bases: object

Rewrites all paths to place them in the specified working directory for TaskWrapper. TaskWrapper is responsible for putting them there and pulling them out.

__init__(working_directory, job_dataset_path_rewriter)[source]
rewrite_dataset_path(dataset, dataset_type)[source]

galaxy.jobs.dynamic_tool_destination module

galaxy.jobs.dynamic_tool_destination.priority_list = set([])

Instantiated to a list of all valid destinations in the job configuration file if run directly to validate configs. Otherwise, remains None. We often check to see if app is None, because if it is then we’ll try using the destination_list instead. -

galaxy.jobs.dynamic_tool_destination.destination_list = set([])

The largest the edit distance can be for a word to be considered A correction for another word.

galaxy.jobs.dynamic_tool_destination.max_edit_dist = 2

List of valid categories that can be expected in the configuration.

exception galaxy.jobs.dynamic_tool_destination.MalformedYMLException[source]

Bases: exceptions.Exception

exception galaxy.jobs.dynamic_tool_destination.ScannerError[source]

Bases: exceptions.Exception

galaxy.jobs.dynamic_tool_destination.get_keys_from_dict(dl, keys_list)[source]

This function builds a list using the keys from nest dictionaries

class galaxy.jobs.dynamic_tool_destination.RuleValidator[source]

Bases: object

This class is the primary facility for validating configs. It’s always called in map_tool_to_destination and it’s called for validating config directly through DynamicToolDestination.py

classmethod validate_rule(rule_type, app, return_bool=False, *args, **kwargs)[source]

This function is responsible for passing each rule to its relevant function.

@type rule_type: str @param rule_type: the current rule’s type

@type return_bool: bool @param return_bool: True when we are only interested in the result of

the validation, and not the validated rule itself.

@rtype: bool, dict (depending on return_bool) @return: validated rule or result of validation (depending on

galaxy.jobs.dynamic_tool_destination.parse_yaml(path='/config/tool_destinations.yml', job_conf_path='/config/job_conf.xml', app=None, test=False, return_bool=False)[source]

Get a yaml file from path and send it to validate_config for validation.

@type path: str @param path: the path to the tool destinations config file

@type job_conf_path: str @param job_conf_path: the path to the job config file

@type test: bool @param test: indicates whether to run in test mode or production mode

@type return_bool: bool @param return_bool: True when we are only interested in the result of the

validation, and not the validated rule itself.

@rtype: bool, dict (depending on return_bool) @return: validated rule or result of validation (depending on return_bool)

galaxy.jobs.dynamic_tool_destination.validate_destination(app, destination, err_message, err_message_contents, return_bool=True)[source]

Validate received destination id.

@type app: @param app: Current app

@type destination: str @param destination: string containing the destination id that is being


@type err_message: str @param err_message: Error message to be formatted with the contents of

err_message_contents upon the event of invalid destination

@type err_message_contents: tuple @param err_message_contents: A tuple of strings to be placed in


@type return_bool: bool @param return_bool: Whether or not the calling function has been told to

return a boolean value or not. Determines whether or not to print ‘Ignoring…’ after error messages.

@rtype: bool @return: True if the destination is valid and False otherwise.

galaxy.jobs.dynamic_tool_destination.validate_config(obj, app=None, return_bool=False)[source]

Validate received config.

@type obj: dict @param obj: the entire contents of the config

@type return_bool: bool @param return_bool: True when we are only interested in the result of the

validation, and not the validated rule itself.

@rtype: bool, dict (depending on return_bool) @return: validated rule or result of validation (depending on return_bool)

galaxy.jobs.dynamic_tool_destination.bytes_to_str(size, unit='YB')[source]

Uses the bi convention: 1024 B = 1 KB since this method primarily has inputs of bytes for RAM

@type size: int @param size: the size in int (bytes) to be converted to str

@rtype: str @return return_str: the resulting string


Uses the bi convention: 1024 B = 1 KB since this method primarily has inputs of bytes for RAM

@type size: str @param size: the size in str to be converted to int (bytes)

@rtype: int @return curr_size: the resulting size converted from str


Uses Mock galaxy for testing or real galaxy for production

@type test: bool @param test: True when being run from a test

galaxy.jobs.dynamic_tool_destination.map_tool_to_destination(job, app, tool, user_email, test=False, path=None, job_conf_path=None)[source]

Dynamically allocate resources

@param job: galaxy job @param app: current app @param tool: current tool

@type test: bool @param test: True when running in test mode

@type path: str @param path: path to tool_destinations.yml

@type job_conf_path: str @param job_conf_path: path to job_conf.xml


returns A list of all destination IDs declared in the job configuration

@type job_config_location: str @param job_config_location: The location of the job config file relative

to the galaxy root directory. If NoneType, defaults to galaxy/config/job_conf.xml, galaxy/config/job_conf.xml.sample_advanced, or galaxy/config/job_conf.xml.sample_basic (first one that exists)

@rtype: list @return: A list of all of the destination IDs declared in the job

configuration file.
galaxy.jobs.dynamic_tool_destination.get_edit_distance(source, target)[source]

returns the edit distance (levenshtein distance) between two strings. code from: en.wikibooks.org/wiki/Algorithm_Implementation/Strings/Levenshtein_distance

@type str1: str @param str1: The first string

@type str2: str @param str2: The second string

@rtype: int @return: The edit distance between str1 and str2

galaxy.jobs.dynamic_tool_destination.get_typo_correction(typo_str, word_set, max_dist)[source]

returns the string in a set that closest matches the input string, as long as the edit distance between them is equal to or smaller than a value, or the words are the same when case is not considered. If there are no appropriate matches, nothing is returned instead.

@type typo_str: str @param typo_str: The string to be compared

@type word_set: set of str @param word_set: The set of strings to compare to

@type max_dist: int @param max_dist: the largest allowed edit distance between

the word and the result. If nothing is within this range, nothing is returned

@rtype: str or NoneType @return: The closest matching string, or None, if no strings being compared to are within max_dist edit distance.

galaxy.jobs.dynamic_tool_destination.verbose = True

list of all valid priorities, inferred from the global default_desinations section of the config

galaxy.jobs.handler module

Galaxy job handler, prepares, runs, tracks, and finishes Galaxy jobs

class galaxy.jobs.handler.JobHandler(app)[source]

Bases: object

Handle the preparation, running, tracking, and finishing of jobs

class galaxy.jobs.handler.JobHandlerQueue(app, dispatcher)[source]

Bases: galaxy.util.monitors.Monitors

Job Handler’s Internal Queue, this is what actually implements waiting for jobs to be runnable and dispatching to a JobRunner.

STOP_SIGNAL = <object object>
__init__(app, dispatcher)[source]

Initializes the Job Handler Queue, creates (unstarted) monitoring thread


Starts the JobHandler’s thread after checking for any unhandled jobs.

job_wrapper(job, use_persisted_destination=False)[source]
increase_running_job_count(user_id, destination_id)[source]
put(job_id, tool_id)[source]

Add a job to the queue (by job identifier)


Attempts to gracefully shut down the worker thread

class galaxy.jobs.handler.JobHandlerStopQueue(app, dispatcher)[source]

Bases: galaxy.util.monitors.Monitors

A queue for jobs which need to be terminated prematurely.

STOP_SIGNAL = <object object>
__init__(app, dispatcher)[source]

Continually iterate the waiting jobs, stop any that are found.


Called repeatedly by monitor to stop jobs.

put(job_id, error_msg=None)[source]

Attempts to gracefully shut down the worker thread

class galaxy.jobs.handler.DefaultJobDispatcher(app)[source]

Bases: object


This is used by the runner mapper (a.k.a. dynamic runner) and recovery methods to have runners convert URLs to destinations.

New-style runner plugin IDs must match the URL’s scheme for this to work.

stop(job, job_wrapper)[source]

Stop the given job. The input variable job may be either a Job or a Task.

recover(job, job_wrapper)[source]

galaxy.jobs.manager module

Top-level Galaxy job manager, moves jobs to handler(s)

class galaxy.jobs.manager.JobManager(app)[source]

Bases: object

Highest level interface to job management.

enqueue(job, tool=None)[source]

Queue a job for execution.

Due to the nature of some handler assignment methods which are wholly DB-based, the enqueue method will flush the job. Callers who create the job typically should not flush the job before handing it off to enqueue(). If a job handler cannot be assigned, :exception:`ToolExecutionError` is raised.


ToolExecutionError – if a handler was unable to be assigned.

returns: str or None – Handler ID, tag, or pool assigned to the job.

stop(job, message=None)[source]

Stop a job that is currently executing.

This can be safely called on jobs that have already terminated.

  • job (Instance of galaxy.model.Job.) – Job to stop.
  • message (str) – Message (if any) to be set on the job and output dataset(s) to explain the reason for stopping.
class galaxy.jobs.manager.NoopManager(*args, **kwargs)[source]

Bases: object

Implements the JobManager interface but does nothing

__init__(*args, **kwargs)[source]
enqueue(*args, **kwargs)[source]
stop(*args, **kwargs)[source]
class galaxy.jobs.manager.NoopHandler(*args, **kwargs)[source]

Bases: object

Implements the JobHandler interface but does nothing

__init__(*args, **kwargs)[source]

galaxy.jobs.mapper module

exception galaxy.jobs.mapper.JobMappingException(failure_message)[source]

Bases: exceptions.Exception

exception galaxy.jobs.mapper.JobNotReadyException(job_state=None, message=None)[source]

Bases: exceptions.Exception

__init__(job_state=None, message=None)[source]
class galaxy.jobs.mapper.JobRunnerMapper(job_wrapper, url_to_destination, job_config)[source]

Bases: object

This class is responsible to managing the mapping of jobs (in the form of job_wrappers) to job runner url strings.

__init__(job_wrapper, url_to_destination, job_config)[source]

cached_job_destination is a public property that is sometimes externally set to short-circuit the mapper, such as during resubmits. get_job_destination will respect that and not run the mapper if so.


Force update of cached_job_destination to mapper determined job destination, overwriting any externally set cached_job_destination

galaxy.jobs.output_checker module

galaxy.jobs.output_checker.check_output(tool, stdout, stderr, tool_exit_code, job)[source]

Check the output of a tool - given the stdout, stderr, and the tool’s exit code, return DETECTED_JOB_STATE.OK if the tool exited succesfully or error type otherwise. No exceptions should be thrown. If this code encounters an exception, it returns OK so that the workflow can continue; otherwise, a bug in this code could halt workflow progress.

Note that, if the tool did not define any exit code handling or any stdio/stderr handling, then it reverts back to previous behavior: if stderr contains anything, then False is returned.

galaxy.jobs.rule_helper module

class galaxy.jobs.rule_helper.RuleHelper(app)[source]

Bases: object

Utility to allow job rules to interface cleanly with the rest of Galaxy and shield them from low-level details of models, metrics, etc….

Currently focus is on figuring out job statistics for a given user, but could interface with other stuff as well.


Job rules can pass this function a job, job_wrapper, or tool and determine if the underlying tool believes it can be containered.

metric_query(select, metric_name, plugin, numeric=True)[source]
should_burst(destination_ids, num_jobs, job_states=None)[source]

Check if the specified destinations destination_ids have at least num_jobs assigned to it - send in job_state as queued to limit this check to number of jobs queued.

See stock_rules for an simple example of using this function - but to get the most out of it - it should probably be used with custom job rules that can respond to the bursting by allocating resources, launching cloud nodes, etc….

choose_one(lst, hash_value=None)[source]

Choose a random value from supplied list. If hash_value is passed in then every request with that same hash_value would produce the same choice from the supplied list.

job_hash(job, hash_by=None)[source]

Produce a reproducible hash for the given job on various criteria - for instance if hash_by is “workflow_invocation,history” - all jobs within the same workflow invocation will receive the same hash - for jobs outside of workflows all jobs within the same history will receive the same hash, other jobs will be hashed on job’s id randomly.

Primarily intended for use with choose_one above - to consistent route or schedule related jobs.

galaxy.jobs.stock_rules module

Stock job ‘dynamic’ rules for use in job_conf.xml - these may cover some simple use cases but will just proxy into functions in rule_helper so similar functionality - but more tailored and composable can be utilized in custom rules.

galaxy.jobs.stock_rules.choose_one(rule_helper, job, destination_ids, hash_by='job')[source]
galaxy.jobs.stock_rules.burst(rule_helper, job, from_destination_ids, to_destination_id, num_jobs, job_states=None)[source]
galaxy.jobs.stock_rules.docker_dispatch(rule_helper, tool, docker_destination_id, default_destination_id)[source]

galaxy.jobs.transfer_manager module

Manage transfers from arbitrary URLs to temporary files. Socket interface for IPC with multiple process configurations.

class galaxy.jobs.transfer_manager.TransferManager(app)[source]

Bases: object

Manage simple data transfers from URLs to temporary locations.

new(path=None, **kwd)[source]

This method blocks, so if invoking the transfer manager ever starts taking too long, we should move it to a thread. However, the transfer_manager will either daemonize or return after submitting to a running daemon, so it should be fairly quick to return.

get_state(transfer_jobs, via_socket=False)[source]