v2.9.0

Added

• Add new runners: winrm-cmd, winrm-ps-cmd and winrm-ps-script.
  The winrm-cmd runner executes Command Prompt commands remotely on Windows hosts using the
  WinRM protocol. The winrm-ps-cmd and winrm-ps-script runners execute PowerShell commands
  and scripts on remote Windows hosts using the WinRM protocol.

  To accompany these new runners, there are two new actions core.winrm_cmd that executes remote
  Command Prompt commands along with core.winrm_ps_cmd that executes remote PowerShell commands.
  (new feature) #1636

  Contributed by Nick Maludy (Encore Technologies).

• Add new ?tags, query param filter to the /v1/actions API endpoint. This query parameter
  allows users to filter out actions based on the tag name . By default, when no filter values are
  provided, all actions are returned. (new feature) #4219

• Add a new standalone standalone st2-pack-install CLI command. This command installs a pack
  (and sets up the pack virtual environment) on the server where it runs. It doesn't register the
  content. It only depends on the Python, git and pip binary and st2common Python package to be
  installed on the system where it runs. It doesn't depend on the database (MongoDB) and message
  bus (RabbitMQ).

  It's primary meant to be used in scenarios where the content (packs) are baked into the base
  container / VM image which is deployed to the cluster.

  Keep in mind that the content itself still needs to be registered with StackStorm at some later
  point when access to RabbitMQ and MongoDB is available by running
  st2ctl reload --register-all. (new feature) #3912 #4256

• Add new /v1/stream/executions/<id>/output[?output_type=all|stdout|stderr] stream API
  endpoint.

  This API endpoint returns event source compatible response format.

  For running executions it returns any output produced so far and any new output as it's produced.
  Once the execution finishes, the connection is automatically closed.

  For completed executions it returns all the output produced by the execution. (new feature)

• Add new core.inject_trigger action for injecting a trigger instance into the system.

  Keep in mind that the trigger which is to be injected must be registered and exist in the system.
  (new feature) #4231 #4259

• Add support for ?include_attributes query param filter to all the content pack resource
  get all (list) API endpoints (actions, rules, trigger, executions, etc.). With this query
  parameter user can control which API model attributes (fields) to receive in the response. In
  situations where user is only interested in a subset of the model attributes, this allows for a
  significantly reduced response size and for a better performance. (new feature) (improvement)
  #4300

• Add new action_sensor.emit_when config option which allows user to specify action status for
  which actiontrigger is emitted. For backward compatibility reasons it defaults to all the action
  completed states. (improvement) #4312 #4315

  Contributed by Shu Sugimoto.

• Improve performance of schedule action execution (POST /v1/executions) API endpoint.

  Performance was improved by reducing the number of duplicated database queries, using atomic
  partial document updates instead of full document updates and by improving database document
  serialization and de-serialization performance. (improvement) #4030 #4331

• Ported existing YAQL and Jinja functions from st2common to Orquesta. (new feature)

• Add error entry in Orquesta workflow result on action execution failure. (improvement)

Changed

• st2 key list command now defaults to --scope=all aka displaying all the datastore values
  (system and current user scoped) . If you only want to display system scoped values (old behavior)
  you can do that by passing --scope=system argument to the st2 key list command
  (st2 key list --scope=system). (improvement) #4221

• The orquesta conductor implemented event based state machines to manage state transition of
  workflow execution. Interfaces to set workflow state and update task on action execution
  completion have changed and calls to those interfaces are changed accordingly. (improvement)

• Change GET /v1/executions/<id>/output API endpoint so it never blocks and returns data
  produced so far for running executions. Behavior for completed executions is the same and didn't
  change - all data produced by the execution is returned in the raw format.

  The streaming (block until execution has finished for running executions) behavior has been moved
  to the new /stream/v1/executions/<id>/output API endpoint.

  This way we are not mixing non-streaming (short lived) and streaming (long lived) connections
  inside a single service (st2api). (improvement)

• Upgrade mongoengine (0.15.3) and pymongo (3.7.1) to the latest stable version. Those
  changes will allow us to support MongoDB 3.6 in the near future.

  New version of mongoengine should also offer better performance when inserting and updating
  larger database objects (e.g. executions). (improvement) #4292

• Trigger parameters and payload schema validation is now enabled by default
  (system.validate_trigger_parameters and system.validate_trigger_payload config options
  now default to True).

  This means that trigger parameters are now validated against the parameters_schema defined on
  the trigger type when creating a rule and trigger payload is validated against payload_schema
  when dispatching a trigger via the sensor or via the webhooks API endpoint.

  This provides a much safer and user-friendly default value. Previously we didn't validate trigger
  payload for custom (non-system) triggers when dispatching a trigger via webhook which meant that
  webhooks API endpoint would silently accept an invalid trigger (e.g. referenced trigger doesn't
  exist in the database or the payload doesn't validate against the payload_schema), but
  TriggerInstanceDB object would never be created because creation failed inside the
  st2rulesengine service. This would make such issues very hard to troubleshoot because only
  way to find out about this failure would be to inspect the st2rulesengine service logs.
  (improvement) #4231

• Improve code metric instrumentation and instrument code and various services with more metrics.
  Also document various exposed metrics. Documentation can be found at
  https://docs.stackstorm.com/latest/reference/metrics.html (improvement) #4310

• Add new metrics.prefix config option. With this option user can specify an optional prefix
  which is prepended to each metric key (name). This comes handy in scenarios where user wants to
  submit metrics from multiple environments / deployments (e.g. testing, staging, dev) to the same
  backend instance. (improvement) #4310

• Improve st2 execution tail CLI command so it also supports Orquesta workflows and arbitrarily
  nested workflows. Also fix the command so it doesn't include data from other unrelated running
  executions. (improvement) #4328

• Change default NGINX configuration to use HTTP 308 redirect, rather than 301, for plaintext requests.
  #4335

• Improve performance of the GET /v1/actions/views/overview API endpoint. (improvement) #4337

Fixed

• Fix an issue with AttributeError: module 'enum' has no attribute 'IntFlag' error which would
  appear when using Python 3 for a particular pack virtual environment and running on RHEL /
  CentOS. (bug fix) #4297

• Fix a bug with action runner throwing an exception and failing to run an action if there was an
  empty pack config inside /opt/stackstorm/configs/. (bug fix) #4325

• Fix action_sensor.enable config option so it works correctly if user sets this option to a
  non-default value of True. (bug fix) #4312 #4315

  Contributed by Shu Sugimoto.

• Update GET /v1/actions/views/entry_point/<action ref> to return correct Content-Type
  response header based on the entry point type / file extension. Previously it would always
  incorrectly return application/json. (improvement) #4327

Deprecated

• The CloudSlang runner is now deprecated. In StackStorm 3.1 it will be removed from the core
  StackStorm codebase. The runner code will be moved to a separate repository, and no longer
  maintained by the core StackStorm team. Users will still be able to install and use this runner,
  but it will require additional steps to install.
• The winexe-based Windows runners are now deprecated. They will be removed in StackStorm 3.1.
  They have been replaced by pywinrm-based Windows runners. See
  https://docs.stackstorm.com/latest/reference/runners.html#winrm-command-runner-winrm-cmd
  for more on using these new runners.

v2.8.1

Added

• Update st2 CLI to inspect COLUMNS environment variable first when determining the
  terminal size. Previously this environment variable was checked second last (after trying to
  retrieve terminal size using various OS specific methods and before falling back to the default
  value).

  This approach is more performant and allows user to easily overwrite the default value or value
  returned by the operating system checks - e.g. by running COLUMNS=200 st2 action list.
  (improvement) #4242

Changed

• Update st2client/setup.py file to dynamically load requirements from
  st2client/requirements.txt file. The code works with pip >= 6.0.0, although using pip 9.0.0
  or higher is strongly recommended. (improvement) #4209

• Update st2 CLI to use a more sensible default terminal size for table formatting purposes if
  we are unable to retrieve terminal size using various system-specific approaches.

  Previously we would fall back to a very unfriendly default of 20 columns for a total terminal
  width. This would cause every table column to wrap and make output impossible / hard to read.
  (improvement) #4242

Fixed

• Fixed a bug where secret: true was not applying to full object and array trees. (bugfix) #4234
  Reported by @jjm

  Contributed by Nick Maludy (Encore Technologies).

• Mark password http-runner parameter as a secret. (bug fix) #4245

  Reported by @daniel-mckenna

v2.8.0

Added

• Orchestra - new StackStorm-native workflow engine. This is currently in beta. (new feature)

• Added metrics for collecting performance and health information about the various ST2 services and functions. (new feature) #4004 #2974

• When running a dev (unstable) release include git revision hash in the output when using st2 --version CLI command. (new feature) #4117

• Update rules engine to also create rule enforcement object when trigger instances fails to match a rule during the rule matching / filtering phase due to an exception in the rule criteria (e.g. invalid Jinja expression, etc.).

• This change increases visibility into rules which didn't match due to an exception. Previously this was only visible / reflected in the rules engine log file. (improvement) #4134

• Add new GET /v1/ruleenforcements/views[/] API endpoints which allow user to retrieve RuleEnforcement objects with the corresponding TriggerInstance and Execution objects. (new feature) #4134

• Add new status field to the RuleEnforcement model. This field can contain the following values - succeeded (trigger instance matched a rule and action execution was triggered successfully), failed (trigger instance matched a rule, but it didn't result in an action execution due to Jinja rendering failure or other exception). (improvement) #4134 #4152

• Add trigger type reference based filtering to the /v1/triggerinstances API endpoint - e.g. /v1/triggerinstances?trigger_type=core.st2.webhook. (new feature) #4151

• Add new --python3 flag to st2 pack install CLI command and python3 parameter to packs.{install,setup_virtualenv} actions. When the value of this parameter is True, it uses python3 binary when creating virtual environment for that pack (based on the value of actionrunner.python3_binary config option).

  Note 1: For this feature to work, Python 3 needs to be installed on the system, virtualenv package installed on the system needs to support Python 3 (it needs to be a recent version) and pack in question needs to support Python 3.

  Note 2: This feature is experimental and opt-in. (new feature) #4016 #3922 #4149

• Add two new Jinja filters - basename (os.path.basename) and dirname (os.path.dirname). #4184

  Contributed by Florian Reisinger (@reisingerf).

Changed

• Update st2 CLI to create the configuration directory and file, and authentication tokens with secure permissions (eg: readable only to owner) #4173

• Refactor the callback module for the post run in runner to be more generic. (improvement)

• Update various Python dependencies to the latest stable versions (gunicorn, gitpython, python-gnupg, tooz, flex). #4110

• Update all the service and script entry points to use /etc/st2/st2.conf as a default value for the config file location.

• This way users don't need to explicitly provide --config-file CLI argument when running various scripts (e.g. st2-track-result, st2-apply-rbac-definitions, etc.) and when they just want to use a default config file. (improvement) #4111

• Update st2 CLI to print a warning if a non-unicode system locale which would prevent StackStorm to function correctly in some scenarios is used. (improvement) #4127 #4120

• Upgrade various internal Python library dependencies to the latest stable versions (kombu, amqp, gitpython, pytz, semver, oslo.utils). (improvement) #4162

• Move from keyczar library to cryptography library for handling symmetric encryption and decryption (secret datastore values).

  Note: This change is fully backward compatible since it just changes the underlying backend and implementation details. The same underlying encryption algorithm is used (AES256 in CBC mode with HMAC signature). (improvement) #4165

Fixed

• Fixed a bug where secrets in pack configs weren't being masked.

  Recently we introduced support for nested objects and arrays. Secret parameters within these nested objects and arrays were not being masked. The fix involves us fully traversing deeply nested objects and arrays and masking out any variables marked as secret. This means we now support pack config JSON schemas with type: object and its corresponding parameters: {} stanza, along with type: array and its corresponding items: {} stanza. We still do NOT support JSON schema combinations that includes the anyOf, allOf, oneOf, and not keywords. (bug fix) #4139

  Contributed by Nick Maludy (Encore Technologies).

• Style clean up to transport queues module and various config modules. (improvement)

• Fixed CLI help for st2 action-alias match and execute. (#4174).

• Fix regression in ?include_attributes query param filter in the /v1/executions API endpoint. (bug fix) #4226

v2.7.2

Changed

• Reduce load on LDAP server and cache user groups response in an in-memory cache when RBAC
  remote LDAP group to local RBAC role synchronization feature is enabled.

  Previously on authentication the code would hit LDAP server multiple times to retrieve user
  groups. With this change, user LDAP groups are only retrieved once upon authentication and
  cached and re-used in-memory by default for 120 seconds.

  This reduces load on LDAP server and improves performance upon regular and concurrent user
  authentication.

  This functionality can be disabled by setting cache_user_groups_response LDAP
  authentication backend kwarg to false.

  Note: This change only affects users which utilize RBAC with remote LDAP groups to local RBAC
  roles synchronization feature enabled. (enterprise) (bug fix) #4103 #4105

Fixed

• Fix an issue (race condition) which would result in not all the remote LDAP groups being
  synchronized with local RBAC roles if a user tried to authenticate with the same auth token
  concurrently in a short time frame.

  Note: This issue only affects users which utilize RBAC with remote LDAP groups to local RBAC
  roles synchronization feature enabled. (enterprise) (bug fix) #4103 #4105

• Fix an issue with some sensors which rely on select.poll() (FileWatch, GithubSensor, etc.)
  stopped working with StackStorm >= 2.7.0.

  StackStorm v2.7.0 inadvertently introduced a change which broke a small set of sensors which
  rely on select.poll() functionality. (bug fix) #4118

• Throw if id CLI argument is not passed to the st2-track-result script. (bug fix) #4115

• Fixed pack config's not properly rendering Jinja expressions within lists. (bugfix) #4121

  Contributed by Nick Maludy (Encore Technologies).

• Fixed pack config rendering error throw meaningful message when a Jinja syntax error is
  encountered. (bugfix) #4123

  Contributed by Nick Maludy (Encore Technologies).

v2.7.1

Changed

• When creating a pack environment during the pack installation, we now pass --no-download flag
  to the virtualenv binary. This way version of pip, wheel and distutils which is enforced by
  virtualenv is used instead of downloading the latest stable versions from PyPi.

  This results in more reproducible pack virtual environments and we also ensure pip 9.0 is used (
  there are some known issues with pip 10.0).

  If for some reason you want to revert to the old behavior, you can do that by passing
  no_download=False parameter to the packs.setup_virtualenv action. #4085

Fixed

• Fix st2 pack search and POST /api/v1/packs/index/search API endpoint so it doesn't
  return internal server error when a single pack search term is provided. (bug fix) #4083

v2.7.0

Added

• Update st2 execution tail command so it supports double nested workflows (workflow ->
  workflow -> execution). Previously, only top-level executions and single nested workflows
  (workflow -> execution) were supported. (improvement) #3962 #3960

• Add support for utf-8 / unicode characters in the pack config files. (improvement) #3980 #3989

  Contributed by @sumkire.

• Added the ability of st2ctl to utilize environment variables from /etc/default/st2ctl
  (for Ubuntu/Debian) and /etc/sysconfig/st2ctl (RHEL/CentOS). This allows
  deployments to override COMPONENTS and ST2_CONF in a global location
  so st2ctl can start/stop/restart selected components and utilize a non-default
  location for st2.conf.
  (new feature) #4027

  Contributed by Nick Maludy (Encore Technologies).

• Add support for new optional content_version runner parameter to the Python and Local Shell
  Script runner. This parameter can contain a git commit hash / tag / branch from a pack git
  repository and runner will ensure this revision of the pack content (Python action / local shell
  script action) is used for a particular action execution.

  Keep in mind that providing this parameter only ensures a particular revision of the pack content
  is used. Python runner virtual environment and dependencies are outside of this scope.

  Note: To be able to utilize this functionality, git version >= 2.5.0 must be installed on the
  system.
  (new feature) #3997

• Update windows runner to correctly handle and use timeout action execution status.
  (improvement) #4047

• Add missing scope, decrypt and encrypt arguments to the datastore management
  related methods on the SensorService class. (improvement) #3895 #4057 #4058

  Reported by @djh2020, @mxmader.

Changed

• Modified RabbitMQ connection error message to make clear that it is an MQ connection issue. #3992

• Additional refactor which makes action runners fully standalone and re-distributable Python
  packages. Also add support for multiple runners (runner modules) inside a single Python package
  and consolidate Python packages from two to one for the following runners: local runners, remote
  runners, windows runners. (improvement) #3999

• Upgrade eventlet library to the latest stable version (0.22.1) (improvement) #4007 #3968

• Increase maximum retry delay for action.retry policy from 5 seconds to 120 seconds. Because
  of the way retries are currently implemented (they are not st2notifier service restart safe),
  long retry delays are not recommended. For more information on this limitation please refer to
  the documentation - https://docs.stackstorm.com/reference/policies.html#retry. #3630 #3637

• Update Python runner so it throws a more user-friendly exception in case Python script tries to
  access a key in self.config dictionary which doesn't exist. (improvement) #4014

• Update various Python dependencies to the latest stable versions (apscheduler, gitpython,
  pymongo, stevedore, paramiko, tooz, flex, webob, prance).

• Refactored mistral runner to support callback from mistral instead of relying on st2resultstracker.
  This reduces the unnecessary traffic and CPU time by querying the mistral API. Included a command to
  manually add a state entry for Mistral workflow execution to recover from any callback failures.
  (improvement)

• Throw a more user-friendly error when writing pack data files to disk and when an invalid file
  path is provided (e.g. path is outside the pack directory, etc.). (improvement) #4039 #4046

• Change the output object returned by Windows runners so it matches the format from the local and
  remote runner.

  Note: This change is backward incompatible - result attribute has been removed (same
  information is available in stdout attribute), exit_code renamed to return_code and
  two new attributes added - succeeded and failed.

  For more information, please refer to the upgrade notes. #4044 #4047

Fixed

• Fix Python runner actions and Argument list too long error when very large parameters are
  passed into the action. The fix utilizes stdin to pass parameters to the Python action wrapper
  process instead of CLI argument list. (bug fix) #1598 #3976

• Fix a regression in POST /v1/webhooks/<webhook name> API endpoint introduced in v2.4.0
  and add back support for arrays. In 2.4.0 support for arrays was inadvertently removed and
  only objects were supported. Keep in mind that this only applies to custom user-defined
  webhooks and system st2 webhook still requires input to be an object (dictionary).
  (bug fix) #3956 #3955

• Fix a bug in the CLI causing st2 execution pause and st2 execution resume
  to not work. (bugfix) #4001

  Contributed by Nick Maludy (Encore Technologies).

• Fixed missing "paused" status option from "st2 execution list" help output. (bugfix) #4037

  Contributed by Ben Hohnke (NTT Communications ICT Solutions)

• Fix "st2 pack install" command so it doesn't require access to pack index (index.stackstorm.org)
  when installing a local pack (pack name starting with "file://"). (bug fix) #3771 #3772

• Fix rules engine so it correctly handles and renders action parameters which contain Jinja
  expressions and default values. (bug fix) #4050 #4050

  Reported by @rakeshrm.

• Make sure observer system role also grants pack_search permission. (bug fix) #4063 #4064

  Reported by @SURAJTHEGREAT.

• Fix st2 webhook get -h which was asking for a name or id as opposed to the URL of the webhook.
  Also, fix st2 webhook list to explicitly add a webhook column. (bugfix) #4048

v2.6.0

https://stackstorm.com/2018/01/25/new-year-new-stackstorm-v2-6-released/

Added

• Add new get_user_info method to action and sensor service. With this method, user can
  retrieve information about the user account which is used to perform datastore operations inside
  the action and sensor service. (new feature) #3831

• Add new /api/v1/user API endpoint. This API endpoint is only available to the authenticated
  users and returns various metadata on the authenticated user (which method did the user use to
  authenticate, under which username the user is authenticated, which RBAC roles are assignment to
  this user in case RBAC is enabled, etc.) (new feature) #3831

• The /api/v1/match_and_execute API endpoint matches a single alias and executes multiple times
  if the alias format has a match_multiple key set to true. Please refer to the
  documentation for usage. #3884

  Contributed by @ahubl-mz.

• Add ability to share common code between python sensors and python actions. You can now place
  common code inside a lib directory inside a pack (with an __init__.py inside lib
  directory to declare it a python package). You can then import the common code in sensors and
  actions. Please refer to documentation for samples and guidelines. #3490

• Add support for password protected sudo to the local and remote runner. Password can be provided
  via the new sudo_password runner parameter. (new feature) #3867

• Add new --tail flag to the st2 run / st2 action execute and st2 execution re-run
  CLI command. When this flag is provided, new execution will automatically be followed and tailed
  after it has been scheduled. (new feature) #3867

• Added flag --auto-dict to st2 run and st2 execution re-run commands. This flag must now
  be specified in order to automatically convert list items to dicts based on presence of colon
  (:) in all of the list items (new feature) #3909

• Allow user to set default log level used by all the Python runner actions by setting
  actionrunner.pythonrunner```` option in st2.conf`` (new feature) #3929

• Update st2client package which is also utilized by the CLI so it also works under Python 3.

  Note: Python 2.7 is only officially supported and tested Python version. Using Python 3 is at
  your own risk - they are likely still many bugs related to Python 3 compatibility. You have been warned.
  (new feature) #3929 #3932

  Contributed by Anthony Shaw.

• Add ?limit=-1 support for the API to fetch full result set (CLI equivalent flag
  --last/-n). Post error message for limit=0 and fix corner case where negative values for
  limit query param were not handled correctly. #3761 #3708 #3735

• Only allow RBAC admins to retrieve all the results at once using ?limit=-1 query param, upate
  the code so api.max_page_size config option only applies to non-admin users, meaning users
  with admin permission can specify arbitrary value for ?limit query param which can also be
  larger than api.max_page_size. (improvement) #3939

• Add new ?include_attributes query param filter to /v1/executions/ API endpoint
  With this filter user can select which fields to include in the response (whitelist approach,
  opposite of the existing ?exclude_attributes filter).

  For example, if you only want to retrieve id and status field, the URL would look like
  this - /v1/executions?include_attributes=id,status. (new feature) #3953 #3858 #3856

Changed

• st2actions.runners.pythonrunner.Action class path for base Python runner actions has been
  deprecated since StackStorm v1.6.0 and will be fully removed in StackStorm v2.7.0. If you have
  any actions still using this path you are encouraged to update them to use
  st2common.runners.base_action.Action path. #3803
• Refactor st2common Python package so it's fully self sustaining and can be used in a
  standalone manner. (improvement) #3803
• Refactor Python action runner so it only depends on st2common Python package (previously it
  also depended on st2actions) and can be used in a standalone mode. Previously pack config and
  and some other parameters were retrieved inside the Python process wrapper, but now they are
  retrieved inside the runner container and passed to the runner. This also makes it easier to add
  support for pack configs to other runners in the future. (improvement) #3803
• Update various Python dependencies to the latest stable versions (kombu, amqp, apscheduler,
  gitpython, pymongo, stevedore, paramiko, prompt-toolkit, flex). #3830
• Mask values in an Inquiry response displayed to the user that were marked as "secret" in the
  inquiry's response schema. #3825
• Real-time action output streaming is now enabled by default. For more information on this
  feature, please refer to the documentation - https://docs.stackstorm.com/latest/reference/action_output_streaming.html.
  You can disable this functionality by setting actionrunner.stream_output config option in
  st2.conf to False and restart the services (sudo st2ctl restart).

Fixed

• Fully fix performance regressions for short Python runner actions introduced in the past and
  partially fixed in #3809. (bug fix) #3803

• Fix 'NameError: name 'cmd' is not defined' error when using linux.service with CentOS systems.
  #3843. Contributed by @shkadov

• Fix bugs with newlines in execution formatter (client) (bug fix) #3872

• Fixed st2ctl status to use better match when checking running process status. #3920

• Removed invalid st2ctl option to re-open Mistral log files. #3920

• Update garbage collection service and st2-purge-executions CLI tool and make deletion more
  efficient. Previously we incorrectly loaded all the execution fields in memory, but there was no
  need for that and now we only retrieve and load id which is the only field we need. #3936

  Reported by @Kevin-vH.

v2.5.1

https://stackstorm.com/2017/12/19/early-christmas-stackstorm-patch-release-2-5-1/

Added

• Add new log_level runner parameter to Python runner. With this parameter, user can control which log messages generated by Python runner actions are output to action stderr. For backward compatibility reasons it defaults to debug.
  This functionality comes handy in situations when an action depends on an external library which logs a lot of information under debug, but you only want to see messages with log level error or higher (or similar). (new feature) #3824

• Add stevedore related metadata to Python package setup.py files for runner packages. This way runners can be installed using pip and dynamically enumerated and loaded using stevedore and corresponding helper functions.

  All runners are now also fully fledged Python packages (previously they were single module
  Python packages which caused various install and distribution related issues when installing
  them via pip) (new feature)

• Add new search rule criteria comparison operator. Please refer to the documentation for
  usage. (new feature) #3833

  Contributed by @ahubl-mz.

• Added flag --auto-dict to st2 run and st2 execution re-run commands. This flag must now
  be specified in order to automatically convert list items to dicts based on presence of colon
  (:) in all of the list items (new feature) #3909

Changed

• Update the output of st2 execution {run,get} CLI command to colorize the value of the
  status attribute (green for succeeded, red for failed, etc. aka the same as for the
  output of st2 execution list command). (improvement) #3810

  Contributed by Nick Maludy (Encore Technologies).

• Update log messages in the datastore service to correctly use DEBUG log level instead of
  AUDIT. #3845

Fixed

• Fix log messages generated by Python runner actions to include the correct action class name.
  Previously they always incorrectly used "ABCMeta" instead of the actual action class name. (bug fix) #3824
• Fix st2 execution tail [last] CLI command so it doesn't throw an exception if there are no executions in the database. (bug fix) #3760 #3802
• Fix a bug with datastore service used inside the Python runner actions not correctly scoping the auth token to the user who triggered the action. Token was incorrectly scoped to api_service user without any permissions. (bug fix) #3823 #3535
• Fix edge case for workflows stuck in running state. When Mistral receives a connection error from the st2 API on requesting action execution, there's a duplicate action execution stuck in requested state. This leads to the st2resultstracker assuming the workflow is still running.
• Fix a regression and a bug with no API validation being performed and API returning 500 instead of 400 status code if user didn't include any request payload (body) when hitting POST and PUT API endpoints where body is mandatory. (bug fix) #3864
• Fix a bug in Python runner which would cause action log messages to be duplicated in action stderr output when utilizing action service / datastore service inside actions. (bug fix) #3893

v2.5.0

https://stackstorm.com/2017/10/26/stackstorm-2-5-hit-streets/

Added

• Add new feature which allows runner action output (stdout and stderr) to be streamed
  and consumed in real-time by using one of the following approaches:

  • /v1/executions/<execution id>/output[?type=stdout/stderr] API endpoint.
  • /v1/stream/ stream endpoint and listening for st2.execution.stdout__create and
    st2.execution.output__create /v1/stream stream API endpoint events.
  • st2 execution tail <execution id> [--type=stdout/stderr] CLI command (underneath it uses
    stream API endpoint).

  Right now this functionality is available for the following runners:

  • local command runner
  • local script runner
  • remote command runner
  • remote script runner
  • python runner

  Note: This feature is still experimental and it's disabled by default (opt-in). To enable it,
  set actionrunner.stream_output config option to True.

  (new feature) #2175 #3657 #3729

• Update st2 role-assignment list RBAC CLI command to include information about where a
  particular assignment comes from (from which local assignment or mapping file). (improvement)
  #3763

• Add support for overlapping RBAC role assignments for assignments via remote LDAP group to
  StackStorm role mappings. This means that the same role can now be granted via multiple RBAC
  mapping files.
  #3763

• Add new Jinja filters from_json_string, from_yaml_string, and jsonpath_query.
  #3763

• Add new "Inquiry" capability, which adds ability to "ask a question", usually in a workflow.
  Create a new runner type: "inquirer" to support this, as well as new API endpoints and
  client commands for interacting with Inquiries

  Contributed by mierdin. #3653

• Added two new rule operators, inside and ninside which allow for the reverse intent of
  the contains and ncontains operators. #3781

  Contributed by @lampwins.

• Allow user to use more expressive regular expressions inside action alias format string by
  allowing them to specify start (^) and end ($) anchors. Previously, those anchors were
  automatically added at the beginning and end of the alias format string. Now they are only added
  if a format string doesn't already contain them. #3789

  Contributed by @ahubl-mz.

• Add new POST /v1/aliasexecution/match_and_execute API endpoint which allows user to
  schedule an execution based on a command string if a matching alias is found in the database.

  This API endpoint is meant to be used with chat bot plugins. It allows them to be simple thin
  wrappers around this API endpoint which send each chat line to this API endpoint and handle the
  response. #3773

• Add several improvements to the installation scripts: They support using proxy servers.
  ~stanley no longer has to be /home/stanley. In addition to the on-screen display, the
  output from the installation script is now logged to a file beginning with st2-install under
  /var/log/st2/. Furthermore, the script handles re-runs better, although it's
  not fully idempotent yet. More improvements are expected in the near future.
  st2-packages: #505, #506, #507, #508, #509, #510, #512, #517.

Fixed

• Fix a bug where sensor watch queues were not deleted after sensor container process was shut
  down. This resulted in spurious queues left behind. This should not have caused performance
  impact but just messes with rabbitmqadmin output and maybe tedious for operators. (bug fix) #3628

  Reported by Igor.

• Make sure all the temporary RabbitMQ queues used by the stream service are deleted once the
  connection to RabbitMQ is closed. Those queues are temporary and unique in nature and new ones
  are created on each service start-up so we need to make sure to correctly clean up old queues. #3746

• Fix cancellation of subworkflow and subchain. Cancel of Mistral workflow or Action Chain is
  cascaded down to subworkflows appropriately. Cancel from tasks in the workflow or chain is
  cascaded up to the parent. (bug fix)

• Fix delays in st2resultstracker on querying workflow status from Mistral. Make sleep time for
  empty queue and no workers configurable. Reduce the default sleep times to 5 seconds. StackStorm
  instances that handle more workflows should consider increasing the query interval for better
  CPU utilization.

• Fix missing type for the parameters with enum in the core st2 packs.(bug fix) #3737

  Reported by Nick Maludy.

• Add missing -h / --help CLI flag to the following execution CLI commands: cancel, pause,
  resume. (bug fix) #3750

• Fix execution cancel and pause CLI commands and make id a required argument. (bug fix) #3750

• Fix st2 role-assignment list CLI command and allow --user, --remote and --role
  arguments to be used together. Previously they were mutually exclusive so it wasn't possible to
  use them together. (bug fix) #3763

• Update default event name whitelist for /v1/stream API endpoint and make sure
  st2.announcement__errbot and other event names starting with st2.announcement__* prefix
  are not filtered out. #3769 (bug fix)

  Reported by Carlos.

• Fix action-alias execute response to show execution id and matching action-alias #3231 (bug fix)
  Reported by Carlos.

• Fix st2 apikey load command to update an existing entry if items in input file contain id
  attribute and item already exists on the server. This way the behavior is consistent with
  st2 key load command and the command is idempotent if each item contains id attribute.
  #3748 #3786

  Reported by Christopher Baklid.

• Don't log MongoDB database password if user specifies URI for database.db_host config
  parameter and that URI also includes a password. Default and a common scenario is specifying
  password as a separate database.password config parameter. #3797

  Reported by Igor Cherkaev.

• Fix POST /v1/actionalias/match API endpoint to correctly return a dictionary instead of an
  array. We had a correct OpenAPI definition for the response, but the code incorrectly returned
  an array instead of a dictionary.

  Note: This is a breaking change so if your code utilizes this API endpoint you need to update
  to treat response as a dictionary and not as an array with a single item. #377

• Partially fix performance overhead and regression for short and simple Python runner actions.
  Full / complete fix will be included in v2.6.0. #3809

Changed

• Minor language and style tidy up of help strings and error messages #3782

v2.4.1

https://stackstorm.com/2017/09/18/quick-one-st2-2-4-1-released/

Fixed

• Fix a bug with /v1/packs/install and /v1/packs/uninstall API endpoints incorrectly using
  system user for scheduled pack install and pack uninstall executions instead of the user which
  performed the API operation.(bug fix) #3693 #3696

  Reported by theuiz.

• Fix mistral callback failure when result contains unicode. (bug fix)

• Fix cancellation of delayed action execution for tasks in workflow. (bug fix)

• Fix timeout of mistral shutdown in systemd service. The fix is done upstream.
  https://review.openstack.org/#/c/499853/ (bug fix)