v1.14.0

This release includes miscellaneous core improvements and bug fixes.

• Miscellaneous improvements:

  • [core] test environment's FSM, handling hooks and fix discovered issues
  • [core] OCTRL-891 propagate pdp_beam_type and pdp sor override to tasks
  • [core] Useless lock is useless

• Bug fixes:

  • [core] Publish correct state when tasks_ is done
  • [occ] avoid a leak in JsonMessage::Deserialize

• Documentation:

  • [docs] Update information on AliECS production deployment
  • [docs] Mention usage of cron in checker script

v1.13.0

This release includes various fixes and improvements, including limiting error payload sizes sent to the GUI, addressing a potential race with Mesos offers handling, and more.

• Mesos offers handling:

  • [core] Improve handling of Mesos resource offers

• Miscellaneous fixes:

  • [core] Set state to DONE in last autoEnv event - stopgap until Kafka switch
  • [core] Limit ODC error string length

• Testing and documentation:

  • test the plugin system
  • Update kafka.md

v1.12.0

This release includes ECS state translation for ODC partitions and devices, support for environment event production in auto-transitioning environments, improvements to critical trait representation, and many miscellaneous improvements and fixes.

• Critical trait:

  • [coconut] Show critical trait of tasks in table
  • [core] Remove TaskClassInfo and add critical trait to ShortTaskInfo
  • [core] Track role criticality and return actual critical trait of tasks
  • [core] Task must return traits for non-basic tasks

• Events in auto-transitioning environments:

  • [core] Emit environment events for Teardown and AutoEnv

• ODC to ECS state translation:

  • [core] Convert ODC device and env states to ECS and publish
  • [core] Move state and transition into their own sm package
  • [core] Account for unlikely ODC state "OK"
  • [core] Consistent variable naming
  • [core] New ODC-ECS state mapping, based on previous ECS state
  • [core] Test ODC-ECS state mapping
  • [core] Move task state test to sm package
  • [core] Strong typing for ODC event payloads
  • [core] Send ODC device IDs as string for JS compatibility
  • [core] Account for ODC state MIXED as INVARIANT
  • [core] Fix test
  • [core] Make test more similar to real world behaviour
  • [core] Fix issue with ECS state reverting to UNKNOWN
  • [core] Ensure we don't overwrite the devices now that it's a pointer

• Miscellaneous:

  • [core] Push CallEvents to aliecs.call topic
  • [core] log the failed JIT workflow and detector, push DPL out only to IL
  • [core] demote message about replying to GetEnvironment call
  • remove generic targets in the operation order doc, since they are not supported
  • spurrious brackets
  • fix links
  • fix detail
  • [occ] Fix deserialization of ConfigEntry with empty value
  • OCTRL-893 [core] always log why the environment goes into ERROR
  • OCTRL-894 [core] a FINISHED/DONE task should have INACTIVE status
  • OCTRL-870 [core] Handle escaped config URIs in DPL commands correctly
  • OCTRL-901 Document the order of actions performed during SOR and EOR

v1.11.0

This release adds support for pushing run information to a new Kafka topic, aliecs.run. It also adds a distinction between start of SOR, end of SOR, start of EOR, and end of EOR (respectively SOSOR, EOSOR, SOEOR, EOEOR), and ensures the lifetime of these values is consistent both within the core and when pushed to controlled tasks. Furthermore, it fixes an issue which prevented FairMQ devices from quitting cleanly from their ERROR state.

• aliecs.run topic and timestamps:

  • [core] Emit run SOSOR/EOSOR/SOEOR/EOEOR events to aliecs.run topic
  • [core] Ensure SOEOR/EOEOR events are pushed in case of kill
  • [core] Push run_end_time_ms at EOR both camelCase and snake_case
  • [core] Clear all old run timestamps on SOSOR
  • [core] Clarify timestamp variables naming

• FairMQ devices behaviour:

  • OCTRL-888 [occ] If we see FairMQ's ERROR state, we should exit

v1.10.0

This release adds support for subdirectories within component configuration prefixes, and removes component configuration entry versioning. Specifically this means that all existing timestamped component configuration entry keys are not transparent any more, and must be referenced explicitly if desired. The recommendation is to redeploy the full component configuration tree.

• Component configuration subpaths:
  • [apricot] allow for arbitrary number of entry subfolders in HTTP service
  • [apricot] test the HTTP handler for Apricot
  • [common] OCTRL-805 allow to group entries in subfolders
  • [common] get rid of the concept of versioning and timestamps
  • [common] add multiple unit tests concerning configuration, include fixes

v1.9.2

This patch release adds snake_case versions of all special SOR parameters (see https://github.com/AliceO2Group/Control/blob/master/docs/handbook/configuration.md#variables-pushed-to-controlled-tasks) to make task compatibility easier between FLP and EPN contexts.

• [core] Push all AliECS-provided run parameters as snake_case+camelCase

v1.9.1

This patch release includes a fix for a core freeze, as well as a feature flag to disable Kafka event production for debugging.

• [core] added internal non mutexed method getParentRolePath to task.go
• [core] added the ability to turn off kafka sending

v1.9.0

This release fixes multiple race conditions in the AliECS core.

• Race conditions:

  • [core] fixed race condition for getTasks in roster.go
  • [core] race condition fix in aggregator.role
  • [core] Added missing mutex around eventStream in environment
  • [core] OCTRL-886 correctly use mutex in environment/manager.go
  • [core] OCTRL-889 proper usage of mutex in environment.go

• Miscellaneous fixes:

  • add a test for GetTasks and GetRoles
  • Bump golang.org/x/net from 0.22.0 to 0.23.0
  • [core] inform, not warn if there is one executor on a node
  • [doc] update the release documentation

v1.8.3

This patch release solves the issue with 824 billion tasks seen in GetEnvironments regardless of the architecture used to build the core, reported in OCTRL-882

• golang bump to 1.22

v1.8.2

This patch release is a fix attempt for OCTRL-881

• [build] added coverage step to makefile
• [core] keeping created kafka writers in global object