Merge pull request #135 from klihub/fixes/nuke-unused-scripts-rules-a…

…nd-defunct-functionality

cleanup: nuke unused scripts, Makefile bits and defunct functionality.

Makefile

@@ -23,13 +23,9 @@ ifdef IMAGE_REPO
     override IMAGE_REPO := $(IMAGE_REPO)/
 endif
 
-# Determine binary version and buildid, and versions for rpm, deb, and tar packages.
+# Determine binary version and buildid.
 BUILD_VERSION := $(shell scripts/build/get-buildid --version --shell=no)
 BUILD_BUILDID := $(shell scripts/build/get-buildid --buildid --shell=no)
-RPM_VERSION   := $(shell scripts/build/get-buildid --rpm --shell=no)
-DEB_VERSION   := $(shell scripts/build/get-buildid --deb --shell=no)
-TAR_VERSION   := $(shell scripts/build/get-buildid --tar --shell=no)
-GOLICENSES_VERSION  ?= v1.6.0
 
 GO_CMD     := go
 GO_BUILD   := $(GO_CMD) build

cmd/plugins/balloons/policy/balloons-policy.go

@@ -30,7 +30,6 @@ import (
 	"github.com/containers/nri-plugins/pkg/resmgr/cache"
 	cpucontrol "github.com/containers/nri-plugins/pkg/resmgr/control/cpu"
 	"github.com/containers/nri-plugins/pkg/resmgr/events"
-	"github.com/containers/nri-plugins/pkg/resmgr/introspect"
 	policy "github.com/containers/nri-plugins/pkg/resmgr/policy"
 	"github.com/containers/nri-plugins/pkg/utils"
 	idset "github.com/intel/goresctrl/pkg/utils"
@@ -283,12 +282,6 @@ func (p *balloons) UpdateResources(c cache.Container) error {
 	return nil
 }
 
-// Rebalance tries to find an optimal allocation of resources for the current containers.
-func (p *balloons) Rebalance() (bool, error) {
-	log.Debug("(not) rebalancing containers...")
-	return false, nil
-}
-
 // HandleEvent handles policy-specific events.
 func (p *balloons) HandleEvent(*events.Policy) (bool, error) {
 	log.Debug("(not) handling event...")
@@ -300,11 +293,6 @@ func (p *balloons) ExportResourceData(c cache.Container) map[string]string {
 	return nil
 }
 
-// Introspect provides data for external introspection.
-func (p *balloons) Introspect(*introspect.State) {
-	return
-}
-
 // GetTopologyZones returns the policy/pool data for 'topology zone' CRDs.
 func (b *balloons) GetTopologyZones() []*policy.TopologyZone {
 	return nil

cmd/plugins/template/policy/template-policy.go

@@ -20,7 +20,6 @@ import (
 	logger "github.com/containers/nri-plugins/pkg/log"
 	"github.com/containers/nri-plugins/pkg/resmgr/cache"
 	"github.com/containers/nri-plugins/pkg/resmgr/events"
-	"github.com/containers/nri-plugins/pkg/resmgr/introspect"
 	policyapi "github.com/containers/nri-plugins/pkg/resmgr/policy"
 )
 
@@ -94,22 +93,12 @@ func (p *policy) UpdateResources(c cache.Container) error {
 	return nil
 }
 
-// Rebalance tries to find an optimal allocation of resources for the current containers.
-func (p *policy) Rebalance() (bool, error) {
-	return true, nil
-}
-
 // HandleEvent handles policy-specific events.
 func (p *policy) HandleEvent(e *events.Policy) (bool, error) {
 	log.Info("received policy event %s.%s with data %v...", e.Source, e.Type, e.Data)
 	return true, nil
 }
 
-// Introspect provides data for external introspection.
-func (p *policy) Introspect(state *introspect.State) {
-	return
-}
-
 // DescribeMetrics generates policy-specific prometheus metrics data descriptors.
 func (p *policy) DescribeMetrics() []*prometheus.Desc {
 	return nil

cmd/plugins/topology-aware/policy/mocks_test.go

@@ -396,18 +396,6 @@ func (m *mockContainer) SetBlockIOClass(string) {
 func (m *mockContainer) GetBlockIOClass() string {
 	panic("unimplemented")
 }
-func (m *mockContainer) SetToptierLimit(int64) {
-	panic("unimplemented")
-}
-func (m *mockContainer) GetToptierLimit() int64 {
-	panic("unimplemented")
-}
-func (m *mockContainer) SetPageMigration(*cache.PageMigrate) {
-	return
-}
-func (m *mockContainer) GetPageMigration() *cache.PageMigrate {
-	return nil
-}
 func (m *mockContainer) GetPending() []string {
 	panic("unimplemented")
 }

cmd/plugins/topology-aware/policy/pools.go

@@ -20,7 +20,6 @@ import (
 
 	"github.com/containers/nri-plugins/pkg/utils/cpuset"
 
-	"github.com/containers/nri-plugins/pkg/kubernetes"
 	"github.com/containers/nri-plugins/pkg/resmgr/cache"
 	system "github.com/containers/nri-plugins/pkg/sysfs"
 	idset "github.com/intel/goresctrl/pkg/utils"
@@ -667,7 +666,6 @@ func (p *policy) applyGrant(grant Grant) {
 	if mems != "" {
 		log.Debug("  => pinning to memory %s", mems)
 		container.SetCpusetMems(mems)
-		p.setDemotionPreferences(container, grant)
 	} else {
 		log.Debug("  => not pinning memory, memory set is empty...")
 	}
@@ -739,34 +737,6 @@ func (p *policy) updateSharedAllocations(grant *Grant) {
 	}
 }
 
-// setDemotionPreferences sets the dynamic demotion preferences a container.
-func (p *policy) setDemotionPreferences(c cache.Container, g Grant) {
-	log.Debug("%s: setting demotion preferences...", c.PrettyName())
-
-	// System containers should not be demoted.
-	if c.GetNamespace() == kubernetes.NamespaceSystem {
-		c.SetPageMigration(nil)
-		return
-	}
-
-	memType := g.GetMemoryNode().GetMemoryType()
-	if memType&memoryDRAM == 0 || memType&memoryPMEM == 0 {
-		c.SetPageMigration(nil)
-		return
-	}
-
-	dram := g.GetMemoryNode().GetMemset(memoryDRAM)
-	pmem := g.GetMemoryNode().GetMemset(memoryPMEM)
-
-	log.Debug("%s: eligible for demotion from %s to %s NUMA node(s)",
-		c.PrettyName(), dram, pmem)
-
-	c.SetPageMigration(&cache.PageMigrate{
-		SourceNodes: dram,
-		TargetNodes: pmem,
-	})
-}
-
 func (p *policy) filterInsufficientResources(req Request, originals []Node) []Node {
 	sufficient := make([]Node, 0)
 

cmd/plugins/topology-aware/policy/topology-aware-policy.go

@@ -18,7 +18,6 @@ import (
 	"errors"
 
 	"github.com/containers/nri-plugins/pkg/utils/cpuset"
-	v1 "k8s.io/api/core/v1"
 	"k8s.io/apimachinery/pkg/api/resource"
 	resapi "k8s.io/apimachinery/pkg/api/resource"
 
@@ -28,7 +27,6 @@ import (
 	"github.com/containers/nri-plugins/pkg/cpuallocator"
 	"github.com/containers/nri-plugins/pkg/resmgr/cache"
 	"github.com/containers/nri-plugins/pkg/resmgr/events"
-	"github.com/containers/nri-plugins/pkg/resmgr/introspect"
 
 	policyapi "github.com/containers/nri-plugins/pkg/resmgr/policy"
 	system "github.com/containers/nri-plugins/pkg/sysfs"
@@ -199,33 +197,6 @@ func (p *policy) UpdateResources(container cache.Container) error {
 	return nil
 }
 
-// Rebalance tries to find an optimal allocation of resources for the current containers.
-func (p *policy) Rebalance() (bool, error) {
-	var errors error
-
-	containers := p.cache.GetContainers()
-	movable := []cache.Container{}
-
-	for _, c := range containers {
-		if c.GetQOSClass() != v1.PodQOSGuaranteed {
-			p.ReleaseResources(c)
-			movable = append(movable, c)
-		}
-	}
-
-	for _, c := range movable {
-		if err := p.AllocateResources(c); err != nil {
-			if errors == nil {
-				errors = err
-			} else {
-				errors = policyError("%v, %v", errors, err)
-			}
-		}
-	}
-
-	return true, errors
-}
-
 // HandleEvent handles policy-specific events.
 func (p *policy) HandleEvent(e *events.Policy) (bool, error) {
 	log.Debug("received policy event %s.%s with data %v...", e.Source, e.Type, e.Data)
@@ -256,45 +227,6 @@ func (p *policy) HandleEvent(e *events.Policy) (bool, error) {
 	return false, nil
 }
 
-// Introspect provides data for external introspection.
-func (p *policy) Introspect(state *introspect.State) {
-	pools := make(map[string]*introspect.Pool, len(p.pools))
-	for _, node := range p.nodes {
-		cpus := node.GetSupply()
-		pool := &introspect.Pool{
-			Name:   node.Name(),
-			CPUs:   cpus.SharableCPUs().Union(cpus.IsolatedCPUs()).String(),
-			Memory: node.GetMemset(memoryAll).String(),
-		}
-		if parent := node.Parent(); !parent.IsNil() {
-			pool.Parent = parent.Name()
-		}
-		if children := node.Children(); len(children) > 0 {
-			pool.Children = make([]string, 0, len(children))
-			for _, c := range children {
-				pool.Children = append(pool.Children, c.Name())
-			}
-		}
-		pools[pool.Name] = pool
-	}
-	state.Pools = pools
-
-	assignments := make(map[string]*introspect.Assignment, len(p.allocations.grants))
-	for _, g := range p.allocations.grants {
-		a := &introspect.Assignment{
-			ContainerID:   g.GetContainer().GetID(),
-			CPUShare:      g.SharedPortion(),
-			ExclusiveCPUs: g.ExclusiveCPUs().Union(g.IsolatedCPUs()).String(),
-			Pool:          g.GetCPUNode().Name(),
-		}
-		if g.SharedPortion() > 0 || a.ExclusiveCPUs == "" {
-			a.SharedCPUs = g.SharedCPUs().String()
-		}
-		assignments[a.ContainerID] = a
-	}
-	state.Assignments = assignments
-}
-
 // DescribeMetrics generates policy-specific prometheus metrics data descriptors.
 func (p *policy) DescribeMetrics() []*prometheus.Desc {
 	return nil

docs/resource-policy/developers-guide/architecture.md

@@ -63,12 +63,8 @@ for:
 
 NRI-RP implements an event processing pipeline. In addition to NRI events,
 it processes a set of other events that are not directly related to or the
-result of NRI requests.
-These events are typically internally generated within NRI-RP. They can be
-the result of changes in the state of some containers or the utilization
-of a shared system resource, which potentially could warrant an attempt to
-rebalance the distribution of resources among containers to bring the system
-closer to an optimal state. Some events can also be generated by policies.
+result of NRI requests. These events are typically internally generated within
+NRI-RP.
 
 The Resource Manager component of NRI-RP implements the basic control
 flow of the processing pipeline. It passes control to all the
@@ -114,18 +110,6 @@ following, based on the event type:
    2. Call policy event handler.
    3. Invoke the controller layer for post-policy processing (same as step 5 for requests).
    4. Release the pipeline lock.
- - For metrics events:
-   1. Perform collection/processing/correlation.
-   2. Engage the processing pipeline lock.
-   3. Update cache objects as/if necessary.
-   4. Request rebalancing as/if necessary.
-   5. Release pipeline lock.
- - For rebalance events:
-   1. Engage the processing pipeline lock.
-   2. Invoke policy layer for rebalancing.
-   3. Invoke the controller layer for post-policy processing (same as step 5 for requests).
-   4. Release the pipeline lock.
-
 
 ### [Cache](tree:/pkg/resmgr/cache/)
 

docs/resource-policy/policy/topology-aware.md

@@ -113,8 +113,6 @@ The `topology-aware` policy has the following features:
         typically found on some special-purpose computing systems
   - cold start
     * pin workload exclusively to PMEM for an initial warm-up period
-  - dynamic page demotion
-    * forcibly migrate read-only and idle container memory pages to PMEM
 
 ## Activating the Policy
 
@@ -359,31 +357,6 @@ In the above example, `container1` would be initially granted only PMEM
 memory controller, but after 60 seconds the DRAM controller would be
 added to the container memset.
 
-## Dynamic Page Demotion
-
-The `topology-aware` policy also supports dynamic page demotion. With dynamic
-demotion enabled, rarely-used pages are periodically moved from DRAM to PMEM
-for those workloads which are assigned to use both DRAM and PMEM memory types.
-The configuration for this feature is done using three configuration keys:
-`DirtyBitScanPeriod`, `PageMovePeriod`, and `PageMoveCount`. All of these
-parameters need to be set to non-zero values in order for dynamic page demotion
-to get enabled. See this configuration file fragment as an example:
-
-```yaml
-policy:
-  Active: topology-aware
-  topology-aware:
-    DirtyBitScanPeriod: 10s
-    PageMovePeriod: 2s
-    PageMoveCount: 1000
-```
-
-In this setup, every pid in every container in every non-system pod
-fulfilling the memory container requirements would have their page ranges
-scanned for non-accessed pages every ten seconds. The result of the scan
-would be fed to a page-moving loop, which would attempt to move 1000 pages
-every two seconds from DRAM to PMEM.
-
 ## Container memory requests and limits
 
 Due to inaccuracies in how `nri-resource-policy` calculates memory requests for