Processes construct their own supervision by WilliamRoebuck · Pull Request #522 · eclipse-score/lifecycle

WilliamRoebuck · 2026-08-24T16:39:44Z

Continuation of the refactoring started in #477. This PR moves alive supervision construction to the ProcessInfoNode constructor.

In addition, it:

Resolves Alive monitor refactoring: Allow components to register their own alive supervision #486
Combines AliveMonitorThread & AliveMonitorImpl
Moves the supervision event buffer into the PHM daemon object
- Previously, launch manager held a reference to the buffer to push supervision deactivation/activation events. With the SupervisionHandle, this is no longer needed
- The buffer is now held by the ObservableEventReader
Corrected check in getTimeForReport
Removed most of supervision_control_client
- Mostly made redundant by the alive monitor's ownership of the buffer

TODO

Simplify/Remove PhmDaemon::construct method. This really just needs to reserve the size of a vector and LM should calculate what size to reserve
Review constness on new methods/parameters (probably most can be made const)
Add brief docs for new methods, update docs where params have changed
Update UTs

github-actions · 2026-08-24T16:41:00Z

License Check Results

🚀 The license check job ran with the Bazel command:

bazel run --lockfile_mode=error //:license-check

Status: ⚠️ Needs Review

Click to expand output

[License Check Output]
Extracting Bazel installation...
Starting local Bazel server (8.7.0) and connecting to it...
INFO: Invocation ID: 532e1479-a65e-45cc-b39a-e3dfba4e28a8
Computing main repo mapping: 
Computing main repo mapping: 
Loading: 
Loading: 0 packages loaded
Loading: 0 packages loaded
Loading: 0 packages loaded
    currently loading: 
Loading: 0 packages loaded
    currently loading: 
Loading: 0 packages loaded
    currently loading: 
Loading: 0 packages loaded
    currently loading: 
Loading: 0 packages loaded
    currently loading: 
Analyzing: target //:license-check (1 packages loaded, 0 targets configured)
Analyzing: target //:license-check (1 packages loaded, 0 targets configured)

Analyzing: target //:license-check (65 packages loaded, 10 targets configured)

Analyzing: target //:license-check (145 packages loaded, 1147 targets configured)

Analyzing: target //:license-check (156 packages loaded, 6447 targets configured)

Analyzing: target //:license-check (158 packages loaded, 9359 targets configured)

Analyzing: target //:license-check (168 packages loaded, 9417 targets configured)

INFO: Analyzed target //:license-check (173 packages loaded, 11555 targets configured).
[7 / 15] [Prepa] Expanding template external/score_tooling+/dash/tool/formatters/_dash_format_converter_stage2_bootstrap.py [for tool]
INFO: Found 1 target...
Target //:license.check.license_check up-to-date:
  bazel-bin/license.check.license_check
  bazel-bin/license.check.license_check.jar
INFO: Elapsed time: 19.834s, Critical Path: 0.25s
INFO: 16 processes: 4 disk cache hit, 12 internal.
INFO: Build completed successfully, 16 total actions
INFO: Running command line: bazel-bin/license.check.license_check ./formatted.txt <args omitted>
usage: org.eclipse.dash.licenses.cli.Main [-batch <int>] [-cd <url>]
       [-confidence <int>] [-ef <url>] [-excludeSources <sources>] [-help] [-lic
       <url>] [-project <shortname>] [-repo <url>] [-review] [-summary <file>]
       [-timeout <seconds>] [-token <token>]

danth

This looks like a good cleanup, just a few small tweaks 😄

…into feature/process-makes-supervision

TimoSteuerwaldETAS · 2026-09-01T14:30:27Z

+    const std::size_t supervised_components)
+    : m_recovery_client(recovery_client), m_config(config)
 {
+    initResult = init(supervised_components);


It doesn't look like that AliveMonitorImpl can recover from initial init error, so why not just move the init() method to the constructor and let it throw exceptions?

Hmm, I think we should avoid throwing exceptions where possible, I'd have to add throw statements to make this throw on some errors I think. I think all the create/construct/init methods in alive monitor should be removed/reworked, but this relates to #488 and is a larger change.

TimoSteuerwaldETAS · 2026-09-01T15:08:47Z

+      buffer_(std::make_shared<SupervisionBufferType>()),
      supervisionManager{std::make_unique<factory::FlatCfgFactory>()},
-      processStateReader{std::move(f_observable_event_receiver)}
+      processStateReader{buffer_}


I am wondering if this variable is named correctly. Technical it is an ObservableEventReader, so observableEventReader_ would be more proper. From a semantic point of view it is a supervisionStateReader_, which I would like even more.

Ah, that one must have missed a rename. I've named it as you suggested cefd51c

TimoSteuerwaldETAS · 2026-09-01T15:11:08Z

    /// @brief For fixed time-step execution during the cyclic execution
    CycleTimer cycleTimer;

+    std::shared_ptr<SupervisionBufferType> buffer_;


Documentation is missing. BTW, the whole naming schema for members is currently a little bit mixed (member_variable vs. member_variable_). I do not think this needs to be fixed in this PR, but we should take time for that. Should I open an issue?

Added a doc in cefd51c. Yeah that's worth opening an issue for I think, we need to decide on a format and stick to it (with a clang-tidy rule)

TimoSteuerwaldETAS · 2026-09-01T15:39:14Z

+    /// @brief Set up alive supervision for the identified process. Alive supervision is not started until the publisher
+    /// is notified.


This took me a while until I understood that this is meant:

Suggested change

/// @brief Set up alive supervision for the identified process. Alive supervision is not started until the publisher

/// is notified.

/// @brief Set up alive supervision for the identified process. Alive supervision is not started until the

/// ISupervisionEventPublisher is notified via reportActivation(..)

However I am also not sure if ISupervisionEventPublisher is the correct word for this interface at all.

Added some more detail d590b6d

TimoSteuerwaldETAS · 2026-09-01T15:51:23Z

+    {
+        SCORE_LANGUAGE_FUTURECPP_ASSERT_DBG_MESSAGE(
+            !supervisionManager.full(), "More alive supervisions than expected were constructed");
+        supervisionManager.constructWorker(id, config, uid, recoveryClient, processStateReader);


Is it okay to ignore the return value here?

Ah, thanks for spotting that, that is a problem! Handled properly in d590b6d

TimoSteuerwaldETAS · 2026-09-01T16:10:06Z

+    const AliveSupervisionConfig& m_config;
+    std::thread alive_monitor_thread_{};
+    std::atomic_bool stop_thread_{false};
+    saf::daemon::EInitCode initResult{saf::daemon::EInitCode::kNotInitialized};


Rename to ìnitResult_, also to not shadow initResultinAliveMonitorImpl::init(..)`

Suggested change

saf::daemon::EInitCode initResult{saf::daemon::EInitCode::kNotInitialized};

saf::daemon::EInitCode initResult_{saf::daemon::EInitCode::kNotInitialized};

Member has now been removed

TimoSteuerwaldETAS · 2026-09-01T16:11:35Z

    OsClock m_osClock{};
-    UptrISupervisionControlReceiver m_observable_event_receiver;
-    const Config& m_config;
+    const AliveSupervisionConfig& m_config;


For new code it would be nice to stick to one naming schema.

Suggested change

const AliveSupervisionConfig& m_config;

const AliveSupervisionConfig& config_;

Done in cefd51c

TimoSteuerwaldETAS · 2026-09-01T16:14:49Z

@@ -44,8 +44,9 @@ class ProcessInfoNode final : public IComponent
    /// @brief Constructs a ProcessInfoNode.
    /// @param config Configuration for the OS process.
    /// @param index The process index within its process group.


Fixed cefd51c

TimoSteuerwaldETAS · 2026-09-01T16:15:03Z

    /// @param index The process index within its process group.
-    /// @param ready_condition Whether this process is considered ready when running or when terminated.
    /// @param process_handling The interfaces used to start, stop and report on the OS process.
+    /// @param supervision_factory Temporary reference to a factory this node can use to construct its supervision if


Fixed cefd51c

TimoSteuerwaldETAS · 2026-09-01T16:25:39Z

This interface does not describe an API which publishes anything. I need to think about what it is actually defining. Something like IProcessState? If you have better ideas, please go ahead.

Renamed to "supervision state reporter", what do you think? 2290950

MaciejKaszynski · 2026-09-02T06:38:04Z

+        LM_LOG_DEBUG() << "Setting up alive supervision for" << identifier_;
+
+        config_.deployment_config.environmental_variables.add(
+            "LCM_ALIVE_INTERFACE_PATH", aliveInterfacePath(identifier_));


Could we remove the env var? Should maybe the constructSupervision just take it as a param?

I don't think I can, the library still need to read the supervision path from somewhere. In #571 the factory will return the path to use though

…into feature/process-makes-supervision

github-project-automation Bot added this to LCM - Lifecycle & Health FT Aug 24, 2026

github-project-automation Bot moved this to Backlog in LCM - Lifecycle & Health FT Aug 24, 2026

WilliamRoebuck requested a deployment to workflow-approval August 24, 2026 16:39 — with GitHub Actions Waiting

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from 663cb89 to 2a5e575 Compare August 27, 2026 16:08

WilliamRoebuck requested a deployment to workflow-approval August 27, 2026 16:08 — with GitHub Actions Waiting

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from 2a5e575 to 5c1f0cc Compare August 28, 2026 08:56

WilliamRoebuck requested a deployment to workflow-approval August 28, 2026 08:56 — with GitHub Actions Waiting

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from 5c1f0cc to 8c2033c Compare August 28, 2026 08:59

WilliamRoebuck requested a deployment to workflow-approval August 28, 2026 08:59 — with GitHub Actions Waiting

WilliamRoebuck marked this pull request as ready for review August 28, 2026 08:59

WilliamRoebuck requested review from FScholPer, MaciejKaszynski, NicolasFussberger, anmittag, antonkri, pawelrutkaq and ramceb as code owners August 28, 2026 08:59

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from 8c2033c to 9e9809f Compare August 28, 2026 09:42

WilliamRoebuck requested a deployment to workflow-approval August 28, 2026 09:43 — with GitHub Actions Waiting

danth suggested changes Aug 28, 2026

View reviewed changes

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from 9e9809f to fdec3cd Compare August 28, 2026 11:45

WilliamRoebuck requested a deployment to workflow-approval August 28, 2026 11:45 — with GitHub Actions Waiting

danth approved these changes Aug 28, 2026

View reviewed changes

WilliamRoebuck added 7 commits August 31, 2026 14:01

Cleanup

d0a26c1

Fix UTs

059727b

Fix wrong member name

3325909

Superficial cleanup

a8f71ca

Move supervision factory into struct

e339ab9

Apply suggestions

f5cc8b4

Fixes after rebase

84780eb

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from f0e336a to 84780eb Compare August 31, 2026 13:16

WilliamRoebuck requested a deployment to workflow-approval August 31, 2026 13:16 — with GitHub Actions Waiting

WilliamRoebuck requested a deployment to workflow-approval September 1, 2026 07:46 — with GitHub Actions Waiting

WilliamRoebuck requested a deployment to workflow-approval September 1, 2026 11:37 — with GitHub Actions Waiting

Merge branch 'main' of https://github.com/eclipse-score/inc_lifecycle …

1a33227

…into feature/process-makes-supervision

WilliamRoebuck force-pushed the feature/process-makes-supervision branch from 32d14e6 to 1a33227 Compare September 1, 2026 15:55

WilliamRoebuck requested a deployment to workflow-approval September 1, 2026 15:55 — with GitHub Actions Waiting

TimoSteuerwaldETAS suggested changes Sep 1, 2026

View reviewed changes

MaciejKaszynski reviewed Sep 2, 2026

View reviewed changes

WilliamRoebuck added 5 commits September 2, 2026 09:41

Initial cleanup

cefd51c

Rename supervision_event_publisher

2290950

Handle failed construction

d590b6d

Extract init method

587f377

Stop alive config invalidation

05759f4

WilliamRoebuck requested a deployment to workflow-approval September 2, 2026 15:06 — with GitHub Actions Waiting

Merge branch 'main' of https://github.com/eclipse-score/inc_lifecycle …

5a021a7

…into feature/process-makes-supervision

WilliamRoebuck requested a deployment to workflow-approval September 2, 2026 15:13 — with GitHub Actions Waiting

		/// @brief Set up alive supervision for the identified process. Alive supervision is not started until the publisher
		/// is notified.

	saf::daemon::EInitCode initResult{saf::daemon::EInitCode::kNotInitialized};
	saf::daemon::EInitCode initResult_{saf::daemon::EInitCode::kNotInitialized};

	const AliveSupervisionConfig& m_config;
	const AliveSupervisionConfig& config_;

Uh oh!

Conversation

WilliamRoebuck commented Aug 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

TODO

Uh oh!

github-actions Bot commented Aug 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

License Check Results

Uh oh!

danth left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

WilliamRoebuck Sep 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

WilliamRoebuck Sep 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

WilliamRoebuck Sep 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

WilliamRoebuck Sep 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

WilliamRoebuck Sep 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

WilliamRoebuck commented Aug 24, 2026 •

edited

Loading

github-actions Bot commented Aug 24, 2026 •

edited

Loading

WilliamRoebuck Sep 2, 2026 •

edited

Loading

WilliamRoebuck Sep 2, 2026 •

edited

Loading

WilliamRoebuck Sep 2, 2026 •

edited

Loading

WilliamRoebuck Sep 2, 2026 •

edited

Loading

WilliamRoebuck Sep 2, 2026 •

edited

Loading