mirror of
https://github.com/ClusterCockpit/cc-metric-collector.git
synced 2024-12-25 15:09:05 +01:00
Push LIKWID collector fix into main (#98)
* InfiniBandCollector: Scale raw readings from octets to bytes * Fix clock frequency coming from LikwidCollector and update docs * Build DEB package for Ubuntu 20.04 for releases * Fix memstat collector with numa_stats option * Remove useless prints from MemstatCollector * Replace ioutils with os and io (#87) * Use lower case for error strings in RocmSmiCollector * move maybe-usable-by-other-cc-components to pkg. Fix all files to use the new paths (#88) * Add collector for monitoring the execution of cc-metric-collector itself (#81) * Add collector to monitor execution of cc-metric-collector itself * Register SelfCollector * Fix import paths for moved packages * Check if at least one CPU with frequency information was detected * Correct type: /proc/stats -> /proc/stat * Update README.md * Run ipmitool asynchron. Improved error handling. * Corrected some typos * Add running average power limit (RAPL) metric collector * Add running average power limit (RAPL) metric collector * Do not mess up with the orignal configuration * * Corrected json config in numastatsMetric.md * Added some debug output to numastatsMetric.go * Fixed computing number of physical packages for non continous physical package IDs (e.g. on Ampere Altra Q80-30) * Fix kernel panic for receiver config with missing receiver type * Add receiver to gather remote IPMI sensor metrics * Added config option to add ipmi-sensors command line options * Add documentaion for IPMI receiver * Update to latest version of included go modules * Add go.mod to App dependency * Try to use common metric tags across hardware vendors * Add IPMI metric: current * remove prefix enumeration like 01-... * Add IPMI receiver example configuration to receivers.json * Minimal formating changes * Add hostlist package * Added tests for hostlist Expand() * Use package hostlist to expand a host list * Use package hostlist to expand a host list * Some servers return "ConsumedPowerWatt":65535 instead of "ConsumedPowerWatt":null * Updated to latest package versions * Do not allow unknown fields in JSON configuration file * Add workflow to customize packages to docs * NFS I/O Stats Collector (#91) * Initial version * Delete values for vanished mount points and comments * Fix for Likwid collector (#95) * Run LIKWID in separate thread and check metric type * Change LIKWID collector documentation to use 'type' instead of 'scope' * Re-initialize LIKWID after one read is missing due to lock toggle * Register cc-metric-collector at Zenodo (#93) * Add initial version of Zenodo project file * Orcid ID added * Update .zenodo.json Co-authored-by: Holger Obermaier <holger.obermaier@kit.edu> * Update ipmiMetric.go * Use latest LIKWID version for builds * Update README.md * Remove development stuff from Makefile * Add Requires(pre) to RPM SPEC file * Use curly brackets in packaging make targets * Fix for LIKWID collector with separate measurement thread and inotify watcher on the LIKWID lock (#97) Co-authored-by: Holger Obermaier <40787752+ho-ob@users.noreply.github.com> Co-authored-by: Holger Obermaier <Holger.Obermaier@kit.edu>
This commit is contained in:
parent
162cce0fda
commit
ff0833c413
32
Makefile
32
Makefile
@ -84,7 +84,7 @@ RPM: scripts/cc-metric-collector.spec
|
||||
@COMMITISH="HEAD"
|
||||
@VERS=$$(git describe --tags $${COMMITISH})
|
||||
@VERS=$${VERS#v}
|
||||
@VERS=$$(echo $$VERS | sed -e s+'-'+'_'+g)
|
||||
@VERS=$$(echo $${VERS} | sed -e s+'-'+'_'+g)
|
||||
@eval $$(rpmspec --query --queryformat "NAME='%{name}' VERSION='%{version}' RELEASE='%{release}' NVR='%{NVR}' NVRA='%{NVRA}'" --define="VERS $${VERS}" "$${SPECFILE}")
|
||||
@PREFIX="$${NAME}-$${VERSION}"
|
||||
@FORMAT="tar.gz"
|
||||
@ -96,10 +96,8 @@ RPM: scripts/cc-metric-collector.spec
|
||||
@if [[ "$${GITHUB_ACTIONS}" == true ]]; then
|
||||
@ RPMFILE="$${RPMDIR}/$${ARCH}/$${NVRA}.rpm"
|
||||
@ SRPMFILE="$${SRPMDIR}/$${NVR}.src.rpm"
|
||||
@ echo "RPM: $${RPMFILE}"
|
||||
@ echo "SRPM: $${SRPMFILE}"
|
||||
@ echo "::set-output name=SRPM::$${SRPMFILE}"
|
||||
@ echo "::set-output name=RPM::$${RPMFILE}"
|
||||
@ echo "SRPM=$${SRPMFILE}" >> $${GITHUB_OUTPUT}
|
||||
@ echo "RPM=$${RPMFILE}" >> $${GITHUB_OUTPUT}
|
||||
@fi
|
||||
|
||||
.PHONY: DEB
|
||||
@ -108,29 +106,25 @@ DEB: scripts/cc-metric-collector.deb.control $(APP)
|
||||
@WORKSPACE=$${PWD}/.dpkgbuild
|
||||
@DEBIANDIR=$${WORKSPACE}/debian
|
||||
@DEBIANBINDIR=$${WORKSPACE}/DEBIAN
|
||||
@mkdir --parents --verbose $$WORKSPACE $$DEBIANBINDIR
|
||||
@mkdir --parents --verbose $${WORKSPACE} $${DEBIANBINDIR}
|
||||
#@mkdir --parents --verbose $$DEBIANDIR
|
||||
@CONTROLFILE="$${BASEDIR}/scripts/cc-metric-collector.deb.control"
|
||||
@COMMITISH="HEAD"
|
||||
@git describe --tags --abbrev=0 $${COMMITISH}
|
||||
@VERS=$$(git describe --tags --abbrev=0 $${COMMITISH})
|
||||
@if [ -z "$$VERS" ]; then VERS=${GITHUB_REF_NAME}; fi
|
||||
@if [ -z "$${VERS}" ]; then VERS=${GITHUB_REF_NAME}; fi
|
||||
@VERS=$${VERS#v}
|
||||
@VERS=$$(echo $$VERS | sed -e s+'-'+'_'+g)
|
||||
@VERS=$$(echo $${VERS} | sed -e s+'-'+'_'+g)
|
||||
@ARCH=$$(uname -m)
|
||||
@ARCH=$$(echo $$ARCH | sed -e s+'_'+'-'+g)
|
||||
@ARCH=$$(echo $${ARCH} | sed -e s+'_'+'-'+g)
|
||||
@if [ "$${ARCH}" = "x86-64" ]; then ARCH=amd64; fi
|
||||
@PREFIX="$${NAME}-$${VERSION}_$${ARCH}"
|
||||
@SIZE_BYTES=$$(du -bcs --exclude=.dpkgbuild "$$WORKSPACE"/ | awk '{print $$1}' | head -1 | sed -e 's/^0\+//')
|
||||
@SIZE="$$(awk -v size="$$SIZE_BYTES" 'BEGIN {print (size/1024)+1}' | awk '{print int($$0)}')"
|
||||
#@sed -e s+"{VERSION}"+"$$VERS"+g -e s+"{INSTALLED_SIZE}"+"$$SIZE"+g -e s+"{ARCH}"+"$$ARCH"+g $$CONTROLFILE > $${DEBIANDIR}/control
|
||||
@echo "Version: $$VERS"
|
||||
@echo "Size: $$SIZE"
|
||||
@echo "Arch: $$ARCH"
|
||||
@sed -e s+"{VERSION}"+"$$VERS"+g -e s+"{INSTALLED_SIZE}"+"$$SIZE"+g -e s+"{ARCH}"+"$$ARCH"+g $$CONTROLFILE > $${DEBIANBINDIR}/control
|
||||
@SIZE_BYTES=$$(du -bcs --exclude=.dpkgbuild "$${WORKSPACE}"/ | awk '{print $$1}' | head -1 | sed -e 's/^0\+//')
|
||||
@SIZE="$$(awk -v size="$${SIZE_BYTES}" 'BEGIN {print (size/1024)+1}' | awk '{print int($$0)}')"
|
||||
@sed -e s+"{VERSION}"+"$${VERS}"+g -e s+"{INSTALLED_SIZE}"+"$${SIZE}"+g -e s+"{ARCH}"+"$${ARCH}"+g $${CONTROLFILE} > $${DEBIANBINDIR}/control
|
||||
@make PREFIX=$${WORKSPACE} install
|
||||
@DEB_FILE="cc-metric-collector_$${VERS}_$${ARCH}.deb"
|
||||
@dpkg-deb -b $${WORKSPACE} "$$DEB_FILE"
|
||||
@dpkg-deb -b $${WORKSPACE} "$${DEB_FILE}"
|
||||
@if [ "$${GITHUB_ACTIONS}" = "true" ]; then
|
||||
@ echo "::set-output name=DEB::$${DEB_FILE}"
|
||||
@ echo "DEB=$${DEB_FILE}" >> $${GITHUB_OUTPUT}
|
||||
@fi
|
||||
@rm -r "$${WORKSPACE}"
|
||||
|
@ -8,6 +8,10 @@ There is a single timer loop that triggers all collectors serially, collects the
|
||||
|
||||
The receiver runs as a go routine side-by-side with the timer loop and asynchronously forwards received metrics to the sink.
|
||||
|
||||
|
||||
[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.7438287.svg)](https://doi.org/10.5281/zenodo.7438287)
|
||||
|
||||
|
||||
# Configuration
|
||||
|
||||
Configuration is implemented using a single json document that is distributed over network and may be persisted as file.
|
||||
|
@ -1,5 +1,5 @@
|
||||
# LIKWID version
|
||||
LIKWID_VERSION = 5.2.1
|
||||
LIKWID_VERSION = 5.2.2
|
||||
LIKWID_INSTALLED_FOLDER=$(shell dirname $(shell which likwid-topology 2>/dev/null) 2>/dev/null)
|
||||
|
||||
LIKWID_FOLDER="$(shell pwd)/likwid"
|
||||
|
@ -15,6 +15,7 @@ import (
|
||||
"math"
|
||||
"os"
|
||||
"os/signal"
|
||||
"os/user"
|
||||
"sort"
|
||||
"strconv"
|
||||
"strings"
|
||||
@ -29,12 +30,14 @@ import (
|
||||
topo "github.com/ClusterCockpit/cc-metric-collector/pkg/ccTopology"
|
||||
"github.com/NVIDIA/go-nvml/pkg/dl"
|
||||
"golang.design/x/thread"
|
||||
fsnotify "gopkg.in/fsnotify.v0"
|
||||
)
|
||||
|
||||
const (
|
||||
LIKWID_LIB_NAME = "liblikwid.so"
|
||||
LIKWID_LIB_DL_FLAGS = dl.RTLD_LAZY | dl.RTLD_GLOBAL
|
||||
LIKWID_DEF_ACCESSMODE = "direct"
|
||||
LIKWID_DEF_LOCKFILE = "/var/run/likwid.lock"
|
||||
)
|
||||
|
||||
type LikwidCollectorMetricConfig struct {
|
||||
@ -68,6 +71,7 @@ type LikwidCollectorConfig struct {
|
||||
AccessMode string `json:"access_mode,omitempty"`
|
||||
DaemonPath string `json:"accessdaemon_path,omitempty"`
|
||||
LibraryPath string `json:"liblikwid_path,omitempty"`
|
||||
LockfilePath string `json:"lockfile_path,omitempty"`
|
||||
}
|
||||
|
||||
type LikwidCollector struct {
|
||||
@ -82,7 +86,7 @@ type LikwidCollector struct {
|
||||
basefreq float64
|
||||
running bool
|
||||
initialized bool
|
||||
needs_reinit bool
|
||||
needs_reinit bool
|
||||
likwidGroups map[C.int]LikwidEventsetConfig
|
||||
lock sync.Mutex
|
||||
measureThread thread.Thread
|
||||
@ -198,6 +202,7 @@ func (m *LikwidCollector) Init(config json.RawMessage) error {
|
||||
m.running = false
|
||||
m.config.AccessMode = LIKWID_DEF_ACCESSMODE
|
||||
m.config.LibraryPath = LIKWID_LIB_NAME
|
||||
m.config.LockfilePath = LIKWID_DEF_LOCKFILE
|
||||
if len(config) > 0 {
|
||||
err := json.Unmarshal(config, &m.config)
|
||||
if err != nil {
|
||||
@ -255,12 +260,16 @@ func (m *LikwidCollector) Init(config json.RawMessage) error {
|
||||
}
|
||||
for _, metric := range evset.Metrics {
|
||||
// Try to evaluate the metric
|
||||
if testLikwidMetricFormula(metric.Calc, params) && checkMetricType(metric.Type) {
|
||||
// Add the computable metric to the parameter list for the global metrics
|
||||
cclog.ComponentDebug(m.name, "Checking", metric.Name)
|
||||
if !checkMetricType(metric.Type) {
|
||||
cclog.ComponentError(m.name, "Metric", metric.Name, "uses invalid type", metric.Type)
|
||||
metric.Calc = ""
|
||||
} else if !testLikwidMetricFormula(metric.Calc, params) {
|
||||
cclog.ComponentError(m.name, "Metric", metric.Name, "cannot be calculated with given counters")
|
||||
metric.Calc = ""
|
||||
} else {
|
||||
globalParams = append(globalParams, metric.Name)
|
||||
totalMetrics++
|
||||
} else {
|
||||
metric.Calc = ""
|
||||
}
|
||||
}
|
||||
} else {
|
||||
@ -270,8 +279,11 @@ func (m *LikwidCollector) Init(config json.RawMessage) error {
|
||||
}
|
||||
for _, metric := range m.config.Metrics {
|
||||
// Try to evaluate the global metric
|
||||
if !testLikwidMetricFormula(metric.Calc, globalParams) {
|
||||
cclog.ComponentError(m.name, "Calculation for metric", metric.Name, "failed")
|
||||
if !checkMetricType(metric.Type) {
|
||||
cclog.ComponentError(m.name, "Metric", metric.Name, "uses invalid type", metric.Type)
|
||||
metric.Calc = ""
|
||||
} else if !testLikwidMetricFormula(metric.Calc, globalParams) {
|
||||
cclog.ComponentError(m.name, "Metric", metric.Name, "cannot be calculated with given counters")
|
||||
metric.Calc = ""
|
||||
} else if !checkMetricType(metric.Type) {
|
||||
cclog.ComponentError(m.name, "Metric", metric.Name, "has invalid type")
|
||||
@ -287,77 +299,194 @@ func (m *LikwidCollector) Init(config json.RawMessage) error {
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
return err
|
||||
}
|
||||
|
||||
ret := C.topology_init()
|
||||
if ret != 0 {
|
||||
err := errors.New("failed to initialize topology module")
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
return err
|
||||
}
|
||||
switch m.config.AccessMode {
|
||||
case "direct":
|
||||
C.HPMmode(0)
|
||||
case "accessdaemon":
|
||||
if len(m.config.DaemonPath) > 0 {
|
||||
p := os.Getenv("PATH")
|
||||
os.Setenv("PATH", m.config.DaemonPath+":"+p)
|
||||
}
|
||||
C.HPMmode(1)
|
||||
for _, c := range m.cpulist {
|
||||
C.HPMaddThread(c)
|
||||
}
|
||||
}
|
||||
m.sock2tid = make(map[int]int)
|
||||
tmp := make([]C.int, 1)
|
||||
for _, sid := range topo.SocketList() {
|
||||
cstr := C.CString(fmt.Sprintf("S%d:0", sid))
|
||||
ret = C.cpustr_to_cpulist(cstr, &tmp[0], 1)
|
||||
if ret > 0 {
|
||||
m.sock2tid[sid] = m.cpu2tid[int(tmp[0])]
|
||||
}
|
||||
C.free(unsafe.Pointer(cstr))
|
||||
}
|
||||
|
||||
m.basefreq = getBaseFreq()
|
||||
m.measureThread = thread.New()
|
||||
m.init = true
|
||||
return nil
|
||||
}
|
||||
|
||||
// take a measurement for 'interval' seconds of event set index 'group'
|
||||
func (m *LikwidCollector) takeMeasurement(evset LikwidEventsetConfig, interval time.Duration) (bool, error) {
|
||||
func (m *LikwidCollector) takeMeasurement(evidx int, evset LikwidEventsetConfig, interval time.Duration) (bool, error) {
|
||||
var ret C.int
|
||||
m.lock.Lock()
|
||||
if m.initialized {
|
||||
ret = C.perfmon_setupCounters(evset.gid)
|
||||
if ret != 0 {
|
||||
var err error = nil
|
||||
var skip bool = false
|
||||
cclog.ComponentDebug(m.name, "Setup returns", ret)
|
||||
if ret == -37 {
|
||||
skip = true
|
||||
var gid C.int = -1
|
||||
sigchan := make(chan os.Signal, 1)
|
||||
watcher, err := fsnotify.NewWatcher()
|
||||
if err != nil {
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
}
|
||||
defer watcher.Close()
|
||||
if len(m.config.LockfilePath) > 0 {
|
||||
info, err := os.Stat(m.config.LockfilePath)
|
||||
if err != nil {
|
||||
return true, err
|
||||
}
|
||||
stat := info.Sys().(*syscall.Stat_t)
|
||||
if stat.Uid != uint32(os.Getuid()) {
|
||||
usr, err := user.LookupId(strconv.FormatUint(uint64(stat.Uid), 10))
|
||||
if err == nil {
|
||||
return true, fmt.Errorf("Access to performance counters locked by %s", usr.Username)
|
||||
} else {
|
||||
err = fmt.Errorf("failed to setup performance group %d", evset.gid)
|
||||
}
|
||||
m.lock.Unlock()
|
||||
return skip, err
|
||||
}
|
||||
m.running = true
|
||||
ret = C.perfmon_startCounters()
|
||||
if ret != 0 {
|
||||
var err error = nil
|
||||
var skip bool = false
|
||||
if ret == -37 {
|
||||
skip = true
|
||||
} else {
|
||||
err = fmt.Errorf("failed to setup performance group %d", evset.gid)
|
||||
}
|
||||
m.lock.Unlock()
|
||||
return skip, err
|
||||
}
|
||||
ret = C.perfmon_readCounters()
|
||||
time.Sleep(interval)
|
||||
m.running = false
|
||||
ret = C.perfmon_stopCounters()
|
||||
if ret != 0 {
|
||||
var err error = nil
|
||||
var skip bool = false
|
||||
if ret == -37 {
|
||||
skip = true
|
||||
} else {
|
||||
err = fmt.Errorf("failed to setup performance group %d", evset.gid)
|
||||
}
|
||||
m.lock.Unlock()
|
||||
return skip, err
|
||||
}
|
||||
m.running = false
|
||||
runtime := float64(C.perfmon_getLastTimeOfGroup(evset.gid))
|
||||
// Go over events and get the results
|
||||
for eidx, counter := range evset.eorder {
|
||||
gctr := C.GoString(counter)
|
||||
for _, tid := range m.cpu2tid {
|
||||
res := C.perfmon_getLastResult(evset.gid, C.int(eidx), C.int(tid))
|
||||
fres := float64(res)
|
||||
if m.config.InvalidToZero && (math.IsNaN(fres) || math.IsInf(fres, 0)) {
|
||||
cclog.ComponentDebug(m.name, "Sanitize", gctr, "to zero")
|
||||
fres = 0.0
|
||||
}
|
||||
evset.results[tid][gctr] = fres
|
||||
return true, fmt.Errorf("Access to performance counters locked by %d", stat.Uid)
|
||||
}
|
||||
}
|
||||
for _, tid := range m.cpu2tid {
|
||||
evset.results[tid]["time"] = runtime
|
||||
err = watcher.Watch(m.config.LockfilePath)
|
||||
if err != nil {
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
}
|
||||
}
|
||||
m.lock.Unlock()
|
||||
m.lock.Lock()
|
||||
defer m.lock.Unlock()
|
||||
select {
|
||||
case e := <-watcher.Event:
|
||||
ret = -1
|
||||
if !e.IsAttrib() {
|
||||
ret = C.perfmon_init(C.int(len(m.cpulist)), &m.cpulist[0])
|
||||
}
|
||||
default:
|
||||
ret = C.perfmon_init(C.int(len(m.cpulist)), &m.cpulist[0])
|
||||
}
|
||||
if ret != 0 {
|
||||
return true, fmt.Errorf("failed to initialize library, error %d", ret)
|
||||
}
|
||||
signal.Notify(sigchan, os.Interrupt)
|
||||
signal.Notify(sigchan, syscall.SIGCHLD)
|
||||
select {
|
||||
case <-sigchan:
|
||||
gid = -1
|
||||
case e := <-watcher.Event:
|
||||
gid = -1
|
||||
if !e.IsAttrib() {
|
||||
gid = C.perfmon_addEventSet(evset.estr)
|
||||
}
|
||||
default:
|
||||
gid = C.perfmon_addEventSet(evset.estr)
|
||||
}
|
||||
if gid < 0 {
|
||||
return true, fmt.Errorf("failed to add events %s, error %d", evset.go_estr, gid)
|
||||
} else {
|
||||
evset.gid = gid
|
||||
//m.likwidGroups[gid] = evset
|
||||
}
|
||||
select {
|
||||
case <-sigchan:
|
||||
ret = -1
|
||||
case e := <-watcher.Event:
|
||||
if !e.IsAttrib() {
|
||||
ret = C.perfmon_setupCounters(gid)
|
||||
}
|
||||
default:
|
||||
ret = C.perfmon_setupCounters(gid)
|
||||
}
|
||||
if ret != 0 {
|
||||
return true, fmt.Errorf("failed to setup events '%s', error %d", evset.go_estr, ret)
|
||||
}
|
||||
select {
|
||||
case <-sigchan:
|
||||
ret = -1
|
||||
case e := <-watcher.Event:
|
||||
if !e.IsAttrib() {
|
||||
ret = C.perfmon_startCounters()
|
||||
}
|
||||
default:
|
||||
ret = C.perfmon_startCounters()
|
||||
}
|
||||
if ret != 0 {
|
||||
return true, fmt.Errorf("failed to start events '%s', error %d", evset.go_estr, ret)
|
||||
}
|
||||
select {
|
||||
case <-sigchan:
|
||||
ret = -1
|
||||
case e := <-watcher.Event:
|
||||
if !e.IsAttrib() {
|
||||
ret = C.perfmon_readCounters()
|
||||
}
|
||||
default:
|
||||
ret = C.perfmon_readCounters()
|
||||
}
|
||||
if ret != 0 {
|
||||
return true, fmt.Errorf("failed to read events '%s', error %d", evset.go_estr, ret)
|
||||
}
|
||||
time.Sleep(interval)
|
||||
select {
|
||||
case <-sigchan:
|
||||
ret = -1
|
||||
case e := <-watcher.Event:
|
||||
if !e.IsAttrib() {
|
||||
ret = C.perfmon_readCounters()
|
||||
}
|
||||
default:
|
||||
ret = C.perfmon_readCounters()
|
||||
}
|
||||
if ret != 0 {
|
||||
return true, fmt.Errorf("failed to read events '%s', error %d", evset.go_estr, ret)
|
||||
}
|
||||
for eidx, counter := range evset.eorder {
|
||||
gctr := C.GoString(counter)
|
||||
for _, tid := range m.cpu2tid {
|
||||
res := C.perfmon_getLastResult(gid, C.int(eidx), C.int(tid))
|
||||
fres := float64(res)
|
||||
if m.config.InvalidToZero && (math.IsNaN(fres) || math.IsInf(fres, 0)) {
|
||||
fres = 0.0
|
||||
}
|
||||
evset.results[tid][gctr] = fres
|
||||
}
|
||||
}
|
||||
for _, tid := range m.cpu2tid {
|
||||
evset.results[tid]["time"] = float64(C.perfmon_getLastTimeOfGroup(gid))
|
||||
}
|
||||
select {
|
||||
case <-sigchan:
|
||||
ret = -1
|
||||
case e := <-watcher.Event:
|
||||
if !e.IsAttrib() {
|
||||
ret = C.perfmon_stopCounters()
|
||||
}
|
||||
default:
|
||||
ret = C.perfmon_stopCounters()
|
||||
}
|
||||
if ret != 0 {
|
||||
return true, fmt.Errorf("failed to stop events '%s', error %d", evset.go_estr, ret)
|
||||
}
|
||||
signal.Stop(sigchan)
|
||||
select {
|
||||
case e := <-watcher.Event:
|
||||
if !e.IsAttrib() {
|
||||
C.perfmon_finalize()
|
||||
}
|
||||
default:
|
||||
C.perfmon_finalize()
|
||||
}
|
||||
return false, nil
|
||||
}
|
||||
|
||||
@ -412,7 +541,7 @@ func (m *LikwidCollector) calcEventsetMetrics(evset LikwidEventsetConfig, interv
|
||||
}
|
||||
|
||||
// Go over the global metrics, derive the value out of the event sets' metric values and send it
|
||||
func (m *LikwidCollector) calcGlobalMetrics(interval time.Duration, output chan lp.CCMetric) error {
|
||||
func (m *LikwidCollector) calcGlobalMetrics(groups []LikwidEventsetConfig, interval time.Duration, output chan lp.CCMetric) error {
|
||||
for _, metric := range m.config.Metrics {
|
||||
scopemap := m.cpu2tid
|
||||
if metric.Type == "socket" {
|
||||
@ -422,7 +551,7 @@ func (m *LikwidCollector) calcGlobalMetrics(interval time.Duration, output chan
|
||||
if tid >= 0 {
|
||||
// Here we generate parameter list
|
||||
params := make(map[string]interface{})
|
||||
for _, evset := range m.likwidGroups {
|
||||
for _, evset := range groups {
|
||||
for mname, mres := range evset.metrics[tid] {
|
||||
params[mname] = mres
|
||||
}
|
||||
@ -436,7 +565,7 @@ func (m *LikwidCollector) calcGlobalMetrics(interval time.Duration, output chan
|
||||
if m.config.InvalidToZero && (math.IsNaN(value) || math.IsInf(value, 0)) {
|
||||
value = 0.0
|
||||
}
|
||||
m.gmresults[tid][metric.Name] = value
|
||||
//m.gmresults[tid][metric.Name] = value
|
||||
// Now we have the result, send it with the proper tags
|
||||
if !math.IsNaN(value) {
|
||||
if metric.Publish {
|
||||
@ -460,203 +589,53 @@ func (m *LikwidCollector) calcGlobalMetrics(interval time.Duration, output chan
|
||||
return nil
|
||||
}
|
||||
|
||||
func (m *LikwidCollector) ReInit() error {
|
||||
C.perfmon_finalize()
|
||||
ret := C.perfmon_init(C.int(len(m.cpulist)), &m.cpulist[0])
|
||||
if ret != 0 {
|
||||
return nil
|
||||
}
|
||||
for i, evset := range m.config.Eventsets {
|
||||
var gid C.int
|
||||
if len(evset.Events) > 0 {
|
||||
//skip := false
|
||||
likwidGroup := genLikwidEventSet(evset)
|
||||
gid = C.perfmon_addEventSet(likwidGroup.estr)
|
||||
if gid >= 0 {
|
||||
likwidGroup.gid = gid
|
||||
likwidGroup.internal = i
|
||||
m.likwidGroups[gid] = likwidGroup
|
||||
|
||||
func (m *LikwidCollector) ReadThread(interval time.Duration, output chan lp.CCMetric) {
|
||||
var err error = nil
|
||||
groups := make([]LikwidEventsetConfig, 0)
|
||||
|
||||
for evidx, evset := range m.config.Eventsets {
|
||||
e := genLikwidEventSet(evset)
|
||||
e.internal = evidx
|
||||
skip := false
|
||||
if !skip {
|
||||
// measure event set 'i' for 'interval' seconds
|
||||
skip, err = m.takeMeasurement(evidx, e, interval)
|
||||
if err != nil {
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
return
|
||||
}
|
||||
}
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
func (m *LikwidCollector) LateInit() error {
|
||||
var ret C.int
|
||||
if m.initialized {
|
||||
return nil
|
||||
}
|
||||
switch m.config.AccessMode {
|
||||
case "direct":
|
||||
C.HPMmode(0)
|
||||
case "accessdaemon":
|
||||
if len(m.config.DaemonPath) > 0 {
|
||||
p := os.Getenv("PATH")
|
||||
os.Setenv("PATH", m.config.DaemonPath+":"+p)
|
||||
}
|
||||
C.HPMmode(1)
|
||||
for _, c := range m.cpulist {
|
||||
C.HPMaddThread(c)
|
||||
if !skip {
|
||||
// read measurements and derive event set metrics
|
||||
m.calcEventsetMetrics(e, interval, output)
|
||||
}
|
||||
groups = append(groups, e)
|
||||
}
|
||||
cclog.ComponentDebug(m.name, "initialize LIKWID topology")
|
||||
ret = C.topology_init()
|
||||
if ret != 0 {
|
||||
err := errors.New("failed to initialize LIKWID topology")
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
return err
|
||||
}
|
||||
|
||||
m.sock2tid = make(map[int]int)
|
||||
tmp := make([]C.int, 1)
|
||||
for _, sid := range topo.SocketList() {
|
||||
cstr := C.CString(fmt.Sprintf("S%d:0", sid))
|
||||
ret = C.cpustr_to_cpulist(cstr, &tmp[0], 1)
|
||||
if ret > 0 {
|
||||
m.sock2tid[sid] = m.cpu2tid[int(tmp[0])]
|
||||
}
|
||||
C.free(unsafe.Pointer(cstr))
|
||||
}
|
||||
|
||||
m.basefreq = getBaseFreq()
|
||||
cclog.ComponentDebug(m.name, "BaseFreq", m.basefreq)
|
||||
|
||||
if m.needs_reinit {
|
||||
m.ReInit()
|
||||
m.needs_reinit = false
|
||||
}
|
||||
|
||||
// cclog.ComponentDebug(m.name, "initialize LIKWID perfmon module")
|
||||
// ret = C.perfmon_init(C.int(len(m.cpulist)), &m.cpulist[0])
|
||||
// if ret != 0 {
|
||||
// var err error = nil
|
||||
// C.topology_finalize()
|
||||
// if ret != -22 {
|
||||
// err = errors.New("failed to initialize LIKWID perfmon")
|
||||
// cclog.ComponentError(m.name, err.Error())
|
||||
// } else {
|
||||
// err = errors.New("access to LIKWID perfmon locked")
|
||||
// }
|
||||
// return err
|
||||
// }
|
||||
|
||||
// // While adding the events, we test the metrics whether they can be computed at all
|
||||
// for i, evset := range m.config.Eventsets {
|
||||
// var gid C.int
|
||||
// if len(evset.Events) > 0 {
|
||||
// //skip := false
|
||||
// likwidGroup := genLikwidEventSet(evset)
|
||||
// // for _, g := range m.likwidGroups {
|
||||
// // if likwidGroup.go_estr == g.go_estr {
|
||||
// // skip = true
|
||||
// // break
|
||||
// // }
|
||||
// // }
|
||||
// // if skip {
|
||||
// // continue
|
||||
// // }
|
||||
// // Now we add the list of events to likwid
|
||||
// gid = C.perfmon_addEventSet(likwidGroup.estr)
|
||||
// if gid >= 0 {
|
||||
// likwidGroup.gid = gid
|
||||
// likwidGroup.internal = i
|
||||
// m.likwidGroups[gid] = likwidGroup
|
||||
// }
|
||||
// } else {
|
||||
// cclog.ComponentError(m.name, "Invalid Likwid eventset config, no events given")
|
||||
// continue
|
||||
// }
|
||||
|
||||
// }
|
||||
|
||||
// If no event set could be added, shut down LikwidCollector
|
||||
if len(m.likwidGroups) == 0 {
|
||||
C.perfmon_finalize()
|
||||
C.topology_finalize()
|
||||
err := errors.New("no LIKWID performance group initialized")
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
return err
|
||||
}
|
||||
sigchan := make(chan os.Signal, 1)
|
||||
signal.Notify(sigchan, syscall.SIGCHLD)
|
||||
signal.Notify(sigchan, os.Interrupt)
|
||||
go func() {
|
||||
<-sigchan
|
||||
|
||||
signal.Stop(sigchan)
|
||||
m.initialized = false
|
||||
}()
|
||||
m.initialized = true
|
||||
return nil
|
||||
// calculate global metrics
|
||||
m.calcGlobalMetrics(groups, interval, output)
|
||||
}
|
||||
|
||||
// main read function taking multiple measurement rounds, each 'interval' seconds long
|
||||
func (m *LikwidCollector) Read(interval time.Duration, output chan lp.CCMetric) {
|
||||
var skip bool = false
|
||||
var err error
|
||||
//var skip bool = false
|
||||
//var err error
|
||||
if !m.init {
|
||||
return
|
||||
}
|
||||
|
||||
m.measureThread.Call(func() {
|
||||
if !m.initialized {
|
||||
m.lock.Lock()
|
||||
err = m.LateInit()
|
||||
if err != nil {
|
||||
m.lock.Unlock()
|
||||
cclog.ComponentError(m.name, "lateinit failed")
|
||||
return
|
||||
}
|
||||
m.initialized = true
|
||||
m.lock.Unlock()
|
||||
skip = true
|
||||
}
|
||||
|
||||
if m.initialized && !skip {
|
||||
time := interval
|
||||
for _, evset := range m.likwidGroups {
|
||||
if !skip {
|
||||
// measure event set 'i' for 'interval' seconds
|
||||
skip, err = m.takeMeasurement(evset, interval)
|
||||
if err != nil {
|
||||
cclog.ComponentError(m.name, err.Error())
|
||||
return
|
||||
}
|
||||
}
|
||||
|
||||
if !skip {
|
||||
// read measurements and derive event set metrics
|
||||
m.calcEventsetMetrics(evset, time, output)
|
||||
}
|
||||
}
|
||||
|
||||
if !skip {
|
||||
// use the event set metrics to derive the global metrics
|
||||
m.calcGlobalMetrics(time, output)
|
||||
}
|
||||
if skip {
|
||||
m.needs_reinit = true
|
||||
m.initialized = false
|
||||
}
|
||||
}
|
||||
m.ReadThread(interval, output)
|
||||
})
|
||||
}
|
||||
|
||||
func (m *LikwidCollector) Close() {
|
||||
if m.init {
|
||||
m.init = false
|
||||
cclog.ComponentDebug(m.name, "Closing ...")
|
||||
m.lock.Lock()
|
||||
if m.initialized {
|
||||
cclog.ComponentDebug(m.name, "Finalize LIKWID perfmon module")
|
||||
C.perfmon_finalize()
|
||||
m.initialized = false
|
||||
}
|
||||
m.measureThread.Terminate()
|
||||
m.initialized = false
|
||||
m.lock.Unlock()
|
||||
cclog.ComponentDebug(m.name, "Finalize LIKWID topology module")
|
||||
C.topology_finalize()
|
||||
|
||||
cclog.ComponentDebug(m.name, "Closing done")
|
||||
}
|
||||
}
|
||||
|
@ -10,6 +10,7 @@ The `likwid` collector is probably the most complicated collector. The LIKWID li
|
||||
"liblikwid_path" : "/path/to/liblikwid.so",
|
||||
"accessdaemon_path" : "/folder/that/contains/likwid-accessD",
|
||||
"access_mode" : "direct or accessdaemon or perf_event",
|
||||
"lockfile_path" : "/var/run/likwid.lock",
|
||||
"eventsets": [
|
||||
{
|
||||
"events" : {
|
||||
@ -49,6 +50,7 @@ Additional options:
|
||||
- `access_mode`: Specify LIKWID access mode: `direct` for direct register access as root user or `accessdaemon`. The access mode `perf_event` is current untested.
|
||||
- `accessdaemon_path`: Folder of the accessDaemon `likwid-accessD` (like `/usr/local/sbin`)
|
||||
- `liblikwid_path`: Location of `liblikwid.so` including file name like `/usr/local/lib/liblikwid.so`
|
||||
- `lockfile_path`: Location of LIKWID's lock file if multiple tools should access the hardware counters. Default `/var/run/likwid.lock`
|
||||
|
||||
### Available metric types
|
||||
|
||||
|
@ -10,6 +10,8 @@ BuildRequires: go-toolset
|
||||
BuildRequires: systemd-rpm-macros
|
||||
# for header downloads
|
||||
BuildRequires: wget
|
||||
# Recommended when using the sysusers_create_package macro
|
||||
Requires(pre): /usr/bin/systemd-sysusers
|
||||
|
||||
Provides: %{name} = %{version}
|
||||
|
||||
|
Loading…
Reference in New Issue
Block a user