v0.21.0

This release brings major improvements to AliECS, including multiple fixes for common error conditions, a refactored environment teardown sequence, a reimplementation of the ODC (EPN subcontrol) client based on the new integration plugin system, and miscellaneous bug fixes.

DEPRECATION NOTICE: starting with AliECS v0.21.0 the o2-aliecs-odc-shim component is deprecated and expected to be removed in the near future. It is still accessible by setting odcshim_enabled to true. The existing variable odc_enabled, which was previously used to enable/disable odc-shim, now controls the new implementation.

• Error handling:

  • [core] Transition configure should return immediately after failing the workflow deployment
  • [core] Log which role failed during deployment
  • [core] Improve output on deployment failure
  • [core] Environment should kill tasks after failed deployment
  • [core] Fix panic close nil channel on teardown
  • [core] Adjust CreateAutoEnvironment to the new Teardown
  • [core] Send event with the correct TaskID
  • [executor] Correct timeout for Kill sequence GetState
  • [executor] Fix PID running check for negative PID (PGID)
  • [executor] Kill leftover tasks when rpc failed to start

• Environment termination and cleanup:

  • [core] Ensure every Teardown API call is forced if necessary
  • [core] Refactor environment.Manager.CreateEnvironment
  • [core] Use TeardownEnvironment to cleanup after a failed deployment

• Integration:

  • [core] New ddsched.EnsureTermination call, to be used as DESTROY hook
  • [core] New ODC integration plugin
  • [core] Sort functions in ODC integration plugin
  • [core] Add EnsureCleanup to ODC functions
  • [core] Concurrently initialize all plugins
  • [core] Reduce integration plugin dial timeouts

• Miscellaneous:

  • [coconut] Print number of FLPs for environment
  • [common] Add ALICE Run 3 System ID mapping
  • [common] Add some non-standard system IDs for convenience
  • [core] Expose number of FLPs per environment
  • [core] Await expressions, optional with trigger
  • [executor] Reduce gRPC dial timeout to 30s thanks to modules v4
  • [misc] Update default configuration parameters
  • [walnut] Fix workflow template schema

v0.20.81

This release brings major improvements to AliECS, including multiple fixes for common error conditions, a refactored environment teardown sequence, a reimplementation of the ODC (EPN subcontrol) client based on the new integration plugin system, and miscellaneous minor fixes.

DEPRECATION NOTICE: starting with AliECS v0.21.0 the o2-aliecs-odc-shim component is deprecated and expected to be removed in the near future. It is still accessible by setting odcshim_enabled to true. The existing variable odc_enabled, which was previously used to enable/disable odc-shim, now controls the new implementation.

• Error handling:

  • [core] Transition configure should return immediately after failing the workflow deployment
  • [core] Log which role failed during deployment
  • [core] Improve output on deployment failure
  • [executor] Correct timeout for Kill sequence GetState
  • [executor] Fix PID running check for negative PID (PGID)

• Environment termination and cleanup:

  • [core] Ensure every Teardown API call is forced if necessary
  • [core] Refactor environment.Manager.CreateEnvironment
  • [core] Use TeardownEnvironment to cleanup after a failed deployment

• Integration:

  • [core] New ddsched.EnsureTermination call, to be used as DESTROY hook
  • [core] New ODC integration plugin
  • [core] Sort functions in ODC integration plugin
  • [core] Add EnsureCleanup to ODC functions
  • [core] Concurrently initialize all plugins
  • [core] Reduce integration plugin dial timeouts

• Miscellaneous:

  • [coconut] Print number of FLPs for environment
  • [common] Add ALICE Run 3 System ID mapping
  • [common] Add some non-standard system IDs for convenience
  • [core] Expose number of FLPs per environment
  • [core] Await expressions, optional with trigger
  • [executor] Reduce gRPC dial timeout to 30s thanks to modules v4
  • [walnut] Fix workflow template schema

v0.20.80

This release brings major improvements to AliECS, including multiple fixes for common error conditions, a refactored environment teardown sequence, a reimplementation of the ODC (EPN subcontrol) client based on the new integration plugin system, and miscellaneous minor fixes.

DEPRECATION NOTICE: starting with AliECS v0.21.0 the o2-aliecs-odc-shim component is deprecated and expected to be removed in the near future. It is still accessible by setting odcshim_enabled to true. The existing variable odc_enabled, which was previously used to enable/disable odc-shim, now controls the new implementation.

• Error handling:

  • [core] Transition configure should return immediately after failing the workflow deployment
  • [core] Log which role failed during deployment
  • [core] Improve output on deployment failure
  • [executor] Correct timeout for Kill sequence GetState
  • [executor] Fix PID running check for negative PID (PGID)

• Environment termination and cleanup:

  • [core] Ensure every Teardown API call is forced if necessary
  • [core] Refactor environment.Manager.CreateEnvironment
  • [core] Use TeardownEnvironment to cleanup after a failed deployment

• Integration:

  • [core] New ddsched.EnsureTermination call, to be used as DESTROY hook
  • [core] New ODC integration plugin
  • [core] Sort functions in ODC integration plugin
  • [core] Add EnsureCleanup to ODC functions
  • [core] Concurrently initialize all plugins
  • [core] Reduce integration plugin dial timeouts

• Miscellaneous:

  • [coconut] Print number of FLPs for environment
  • [common] Add ALICE Run 3 System ID mapping
  • [common] Add some non-standard system IDs for convenience
  • [core] Expose number of FLPs per environment
  • [executor] Reduce gRPC dial timeout to 30s thanks to modules v4
  • [walnut] Fix workflow template schema

v0.20.7

This patch release brings several fixes, mainly concerning integration, and to facilitate operations at LHC Point 2.

[coconut] Set maximum incoming gRPC payload size to 10MB
[core] Log when DD scheduler connected
[core] Disable system proxy by default for Mesos, DCS, DDscheduler
[executor] Never use any proxy when connecting to Mesos master

v0.20.6

This patch release fixes a race in the task termination procedure.

• [core] Fix kill tasks deadline

v0.20.5

This patch release fixes an issue that could cause an executor crash when force destroying an environment derived from a workflow template that includes hooks.

• [executor] Ensure hooks can always terminate safely

v0.20.4

This patch release brings an urgent update to the DD scheduler interface, as well as a fix for a rare crash caused by an invalid workflow template structure.

• [core] Bump DD scheduler Protobuf interface
• [core] Prevent crash on empty defaults/vars maps

v0.20.3

This patch release fixes an unusual build issue with certain dependencies of the template system.

• [build] Update dependencies

v0.20.2

This patch release adds a DataDistribution scheduler plugin, as well as improvements to integration plugin loading and better coconut role query output.

• [apricot] Better init output
• [core] DD scheduler integration plugin
• [core][coconut] Display consolidated variable stack in role query
• [core] Fix integration plugin loading

v0.20.1

This is the first patch release in the v0.20 series, it brings a single fix to restore access to Consul-provided defaults and vars.

• [apricot] Fix defaults/vars path