CM launcher memory optimization by mykola-kobets-epam · Pull Request #507 · aosedge/aos_core_lib_cpp

mykola-kobets-epam · 2026-02-10T03:06:07Z

No description provided.

Copilot

Pull request overview

This PR refactors the CM launcher’s scheduling/status flow to reduce memory usage by removing per-node stored instance lists and instead deriving start/stop deltas from centrally tracked running statuses, while also persisting SM-facing aos::InstanceInfo inside each Instance.

Changes:

Move SM instance data (aos::InstanceInfo) into Instance (mSMInfo) and adjust scheduling/network setup to populate it.
Rework NodeManager/Node send/resend logic to operate on provided “scheduled instances” + “running statuses” rather than node-local cached instance arrays; add a shared node allocator.
Update launcher/balancer/instance manager APIs to pass around explicit instance collections and track running statuses in InstanceManager; update tests accordingly (manifest digest/version propagation).

Reviewed changes

Copilot reviewed 15 out of 15 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
src/core/cm/launcher/tests/stubs/instancerunnerstub.hpp	Test stub now propagates manifest digest/version into reported statuses.
src/core/cm/launcher/tests/launcher.cpp	Updates tests/helpers to compute and assert manifest digests; adjusts runner init and expected status notifications.
src/core/cm/launcher/nodemanager.hpp	Renames/repurposes APIs (load SM data, notify node status received) and changes send/resend signatures; adds shared allocator.
src/core/cm/launcher/nodemanager.cpp	Implements the new node init + send/resend + load-SM-data flow using the new APIs/allocator.
src/core/cm/launcher/nodeitf.hpp	Removes `ScheduleInstance` from the node interface.
src/core/cm/launcher/node.hpp	Changes init signature to accept allocator; removes per-node stored instance arrays; updates send/resend signatures.
src/core/cm/launcher/node.cpp	Implements delta computation from running statuses + scheduled instances; adds filtering helper; simplifies stop conversion.
src/core/cm/launcher/launcher.hpp	Replaces “schedule” helpers with “create requested instances” helpers; increases launcher allocator sizing.
src/core/cm/launcher/launcher.cpp	Adjusts lock ordering, instance creation flow, resend calls, and startup loading of SM data for stored active instances.
src/core/cm/launcher/instancemanager.hpp	Adds running-instances tracking + new create/schedule APIs and helpers.
src/core/cm/launcher/instancemanager.cpp	Implements running-status tracking and new create/schedule flow; updates storage-load behavior.
src/core/cm/launcher/instance.hpp	Adds `GetSMInfo()`, embeds `mSMInfo`, and splits network param ops into `Instance` methods; updates schedule signature.
src/core/cm/launcher/instance.cpp	Populates `mSMInfo` during schedule; moves network param setup/removal into instance implementations.
src/core/cm/launcher/balancer.hpp	Changes `RunInstances` signature to accept explicit instance list; renames load method; updates balancing helpers.
src/core/cm/launcher/balancer.cpp	Wires new instance-list-driven balancing, new scheduling API, and updated network param calls.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-02-10T10:10:58Z

    }

    instance = FindCachedInstance(id);
    if (instance) {


FindReadyInstance() returns an instance from mCachedInstances without removing it from the cache. Later, when that same SharedPtr is scheduled and SubmitScheduledInstances() promotes mScheduledInstances to mActiveInstances, the instance can end up in both mActiveInstances and mCachedInstances. This can lead to incorrect behavior (e.g., RemoveOutdatedInstances() deleting an active instance via the cached list) and duplicated bookkeeping. Consider removing the instance from mCachedInstances when it is selected for scheduling (e.g., in FindReadyInstance() when it finds a cached instance, or in ScheduleInstance(...) before adding to mScheduledInstances).

Suggested change

if (instance) {

if (instance) {

// The instance is transitioning out of the cache; remove it from mCachedInstances

for (auto it = mCachedInstances.begin(); it != mCachedInstances.end(); ++it) {

if (*it == instance) {

mCachedInstances.erase(it);

break;

}

}

removed scheduled instance from cached in SubmitScheduledInstances

Copilot · 2026-02-10T10:10:59Z

+RetWithError<SharedPtr<Instance>> InstanceManager::CreateInstance(
+    const InstanceIdent& id, const RunInstanceRequest& request)
 {
+    auto instance = FindReadyInstance(id);


CreateInstance() calls FindReadyInstance(id) twice and also leaves an unused local (auto instance = FindReadyInstance(id);). This is redundant and may trigger unused-variable warnings depending on build flags. Consider removing the first assignment and keeping a single lookup (e.g., the init-statement form in the if).

Suggested change

auto instance = FindReadyInstance(id);

removed duplicate

Copilot · 2026-02-10T10:10:59Z

    auto nodes        = MakeUnique<StaticArray<Node*, cMaxNumNodes>>(&mAllocator);
    auto instanceInfo = MakeUnique<aos::InstanceInfo>(&mAllocator);


Balancer::ScheduleInstance() allocates instanceInfo but no longer uses it after the refactor to InstanceManager::ScheduleInstance(...). This can produce an unused-variable warning (and fail the build if warnings are treated as errors). Consider removing instanceInfo (and any related allocator sizing if it was only needed for that).

Suggested change

auto nodes = MakeUnique<StaticArray<Node*, cMaxNumNodes>>(&mAllocator);

auto instanceInfo = MakeUnique<aos::InstanceInfo>(&mAllocator);

auto nodes = MakeUnique<StaticArray<Node*, cMaxNumNodes>>(&mAllocator);

removed unused instanceInfo

Copilot · 2026-02-10T10:10:59Z

+Error Balancer::PerformPolicyBalancing(Array<SharedPtr<Instance>>& instances)
 {
    auto imageIndex   = MakeUnique<oci::ImageIndex>(&mAllocator);
    auto instanceInfo = MakeUnique<aos::InstanceInfo>(&mAllocator);


PerformPolicyBalancing() still allocates instanceInfo but doesn't use it anymore. This can trigger unused-variable warnings and should be removed.

Suggested change

auto instanceInfo = MakeUnique<aos::InstanceInfo>(&mAllocator);

removed unused instanceInfo

al1img · 2026-02-10T09:45:08Z


 private:
    static constexpr auto cStatusUpdateTimeout = Time::cMinutes * 10;
+


remove empty line and line 145

al1img · 2026-02-10T09:52:39Z

        return AOS_ERROR_WRAP(err);
    }

+    auto loadErr


if (auto err =

I prefer "out of if clause" initialization for long expressions.

al1img · 2026-02-10T09:53:32Z

        if (!isStashed) {
            if (auto err = instance->Cache(); !err.IsNone()) {
-                return AOS_ERROR_WRAP(err);
+                const auto& id = instance->GetInfo().mInstanceIdent;


Why do you need an intermediate var here?

I use aliases to shorten expressions

al1img · 2026-02-10T09:53:38Z

 {
    if (auto err = instance->Cache(true); !err.IsNone()) {
-        return AOS_ERROR_WRAP(err);
+        const auto& id = instance->GetInfo().mInstanceIdent;


I use aliases to shorten expressions

al1img · 2026-02-10T09:55:06Z

+    virtual Error Schedule(NodeItf& node, const String& runtimeID) = 0;
+
+    /**
+     * Setup network parameters.


al1img · 2026-02-10T09:55:13Z

+    virtual Error PrepareNetworkParams(bool onlyExposedPorts) = 0;
+
+    /**
+     * Remove network parameters.


al1img · 2026-02-10T09:55:21Z

+    Error Schedule(NodeItf& node, const String& runtimeID) override;
+
+    /**
+     * Prepare network parameters.


al1img · 2026-02-10T09:55:26Z

+    Error PrepareNetworkParams(bool onlyExposedPorts) override;
+
+    /**
+     * Remove network parameters.


al1img · 2026-02-10T09:55:34Z

+    Error Schedule(NodeItf& node, const String& runtimeID) override;
+
+    /**
+     * Prepare network parameters.


al1img · 2026-02-10T09:55:38Z

+    Error PrepareNetworkParams(bool onlyExposedPorts) override;
+
+    /**
+     * Remove network parameters.


MykolaSuperman · 2026-02-10T09:40:59Z

+        {
+            assert(mIt != mEnd);
+
+            Iterator tmp = *this;


i suppose the beheviour shoud be same as in operator++()

it's a post increment operator i++,
returns incremented copy

MykolaSuperman · 2026-02-10T09:57:53Z

+    const InstanceIdent& id, const RunInstanceRequest& request)
 {
+    auto instance = FindReadyInstance(id);
+    if (auto instance = FindReadyInstance(id); instance) {


why we two times call FindReadyInstance ?

yes, that's a mistake, removed duplicate

codecov · 2026-02-10T11:34:41Z

Codecov Report

❌ Patch coverage is 87.41722% with 38 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.19%. Comparing base (e23272c) to head (628ce86).
⚠️ Report is 11 commits behind head on feature_unification.

Files with missing lines	Patch %	Lines
src/core/cm/launcher/launcher.cpp	72.54%	14 Missing ⚠️
src/core/cm/launcher/instancemanager.cpp	80.00%	11 Missing ⚠️
src/core/cm/launcher/balancer.cpp	85.71%	5 Missing ⚠️
src/core/cm/launcher/instance.cpp	93.47%	3 Missing ⚠️
src/core/cm/launcher/node.cpp	95.00%	3 Missing ⚠️
src/core/cm/launcher/nodemanager.cpp	84.61%	2 Missing ⚠️

Additional details and impacted files

@@                   Coverage Diff                   @@
##           feature_unification     #507      +/-   ##
=======================================================
- Coverage                85.20%   85.19%   -0.02%     
=======================================================
  Files                      308      308              
  Lines                    27154    27181      +27     
  Branches                  3669     3663       -6     
=======================================================
+ Hits                     23136    23156      +20     
- Misses                    4018     4025       +7

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

al1img

Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com>

MykolaSuperman

Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

sonarqubecloud · 2026-02-10T12:30:27Z

Quality Gate failed

Failed conditions
49.7% Coverage on New Code (required ≥ 80%)

See analysis details on SonarQube Cloud

al1img requested a review from Copilot February 10, 2026 10:04

Copilot started reviewing on behalf of al1img February 10, 2026 10:05 View session

Copilot AI reviewed Feb 10, 2026

View reviewed changes

al1img reviewed Feb 10, 2026

View reviewed changes

MykolaSuperman reviewed Feb 10, 2026

View reviewed changes

mykola-kobets-epam force-pushed the integration-fixes branch from 871bedd to 72e99c3 Compare February 10, 2026 11:21

al1img approved these changes Feb 10, 2026

View reviewed changes

MykolaSuperman approved these changes Feb 10, 2026

View reviewed changes

mykola-kobets-epam added 8 commits February 10, 2026 14:11

cm: launcher: don't return error if instance cache failed

5dc6c30

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: move aos::InstanceInfo into Instance class

04142e5

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: remove internal instance arrays from node

51bbc8e

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: adapt nodemanager interface to instance arrays relocation

bd18121

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: extend instancemanager interface

9a4bb6b

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: adapt balancer to interface changes

cfd5f74

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: adapt launcher implementation

32486a3

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

cm: launcher: update launcher unit tests

628ce86

Signed-off-by: Mykola Kobets <mykola_kobets@epam.com> Reviewed-by: Oleksandr Grytsov <oleksandr_grytsov@epam.com> Reviewed-by: Mykola Solianko <mykola_solianko@epam.com>

mykola-kobets-epam force-pushed the integration-fixes branch from 72e99c3 to 628ce86 Compare February 10, 2026 12:11

al1img merged commit 3bfeb08 into aosedge:feature_unification Feb 10, 2026
4 of 5 checks passed

-    if (instance) {
+    if (instance) {
+        // The instance is transitioning out of the cache; remove it from mCachedInstances
+        for (auto it = mCachedInstances.begin(); it != mCachedInstances.end(); ++it) {
+            if (*it == instance) {
+                mCachedInstances.erase(it);
+                break;
+            }
+        }

		auto nodes = MakeUnique<StaticArray<Node*, cMaxNumNodes>>(&mAllocator);
		auto instanceInfo = MakeUnique<aos::InstanceInfo>(&mAllocator);


		private:
		static constexpr auto cStatusUpdateTimeout = Time::cMinutes * 10;

Uh oh!

Conversation

mykola-kobets-epam commented Feb 10, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

mykola-kobets-epam Feb 10, 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!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

mykola-kobets-epam Feb 10, 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!

Copilot AI Feb 10, 2026

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Copilot AI Feb 10, 2026

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!

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!

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!

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!

codecov Bot commented Feb 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

mykola-kobets-epam Feb 10, 2026 •

edited

Loading

mykola-kobets-epam Feb 10, 2026 •

edited

Loading

codecov Bot commented Feb 10, 2026 •

edited

Loading