Merge branch 'develop' into main

.github/workflows/AlmaLinux.yml

@@ -3,8 +3,11 @@
 # Workflow name
 name: AlmaLinux 8.5 RPM build
 
-# Run on event push
-on: push
+# Run on tag push
+on:
+ push:
+  tags:
+  - '**'
 
 jobs:
 

.github/workflows/RedHatUniversalBaseImage.yml

@@ -3,8 +3,11 @@
 # Workflow name
 name: Red Hat Universal Base Image 8 RPM build
 
-# Run on event push
-on: push
+# Run on tag push
+on:
+ push:
+  tags:
+  - '**'
 
 jobs:
 

Makefile

@@ -1,5 +1,5 @@
 APP = cc-metric-collector
-GOSRC_APP        := metric-collector.go
+GOSRC_APP        := cc-metric-collector.go
 GOSRC_COLLECTORS := $(wildcard collectors/*.go)
 GOSRC_SINKS      := $(wildcard sinks/*.go)
 GOSRC_RECEIVERS  := $(wildcard receivers/*.go)

collectors/likwidMetric.go

@@ -2,7 +2,7 @@ package collectors
 
 /*
 #cgo CFLAGS: -I./likwid
-#cgo LDFLAGS: -L./likwid -llikwid -llikwid-hwloc -lm -Wl,--unresolved-symbols=ignore-in-object-files
+#cgo LDFLAGS: -Wl,--unresolved-symbols=ignore-in-object-files
 #include <stdlib.h>
 #include <likwid.h>
 */
@@ -71,8 +71,9 @@ func GetAllMetricScopes() []MetricScope {
 }
 
 const (
-	LIKWID_LIB_NAME     = "liblikwid.so"
-	LIKWID_LIB_DL_FLAGS = dl.RTLD_LAZY | dl.RTLD_GLOBAL
+	LIKWID_LIB_NAME       = "liblikwid.so"
+	LIKWID_LIB_DL_FLAGS   = dl.RTLD_LAZY | dl.RTLD_GLOBAL
+	LIKWID_DEF_ACCESSMODE = "direct"
 )
 
 type LikwidCollectorMetricConfig struct {
@@ -95,6 +96,8 @@ type LikwidCollectorConfig struct {
 	Metrics        []LikwidCollectorMetricConfig   `json:"globalmetrics,omitempty"`
 	ForceOverwrite bool                            `json:"force_overwrite,omitempty"`
 	InvalidToZero  bool                            `json:"invalid_to_zero,omitempty"`
+	AccessMode     string                          `json:"access_mode,omitempty"`
+	DaemonPath     string                          `json:"accessdaemon_path,omitempty"`
 }
 
 type LikwidCollector struct {
@@ -260,6 +263,7 @@ func (m *LikwidCollector) getResponsiblities() map[MetricScope]map[int]int {
 func (m *LikwidCollector) Init(config json.RawMessage) error {
 	var ret C.int
 	m.name = "LikwidCollector"
+	m.config.AccessMode = LIKWID_DEF_ACCESSMODE
 	if len(config) > 0 {
 		err := json.Unmarshal(config, &m.config)
 		if err != nil {
@@ -270,6 +274,11 @@ func (m *LikwidCollector) Init(config json.RawMessage) error {
 	if lib == nil {
 		return fmt.Errorf("error instantiating DynamicLibrary for %s", LIKWID_LIB_NAME)
 	}
+	err := lib.Open()
+	if err != nil {
+		return fmt.Errorf("error opening %s: %v", LIKWID_LIB_NAME, err)
+	}
+
 	if m.config.ForceOverwrite {
 		cclog.ComponentDebug(m.name, "Set LIKWID_FORCE=1")
 		os.Setenv("LIKWID_FORCE", "1")
@@ -301,6 +310,16 @@ func (m *LikwidCollector) Init(config json.RawMessage) error {
 	m.initGranularity()
 	// Generate map for MetricScope -> scope_id (like socket id) -> responsible id (offset in cpulist)
 	m.scopeRespTids = m.getResponsiblities()
+	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)
+	}
 
 	cclog.ComponentDebug(m.name, "initialize LIKWID perfmon module")
 	ret = C.perfmon_init(C.int(len(m.cpulist)), &m.cpulist[0])

collectors/likwidMetric.md

@@ -8,6 +8,8 @@ The `likwid` configuration consists of two parts, the "eventsets" and "globalmet
 - The global metrics are metrics which require data from all event set measurements to be derived. The inputs are the metrics in the event sets. Similar to the metrics in the event sets, the global metrics are defined by a name, a formula, a scope and a publish flag. See event set metrics for details. The only difference is that there is no access to the raw event measurements anymore but only to the metrics. So, the idea is to derive a metric in the "eventsets" section and reuse it in the "globalmetrics" part. If you need a metric only for deriving the global metrics, disable forwarding of the event set metrics. **Be aware** that the combination might be misleading because the "behavior" of a metric changes over time and the multiple measurements might count different computing phases.
 
 Additional options:
+- `access_mode` : Method to use for hardware performance monitoring (`direct` access as root user, `accessdaemon` for the daemon mode)
+- `accessdaemon_path`: Folder with the access daemon `likwid-accessD`, commonly `$LIKWID_INSTALL_LOC/sbin`
 - `force_overwrite`: Same as setting `LIKWID_FORCE=1`. In case counters are already in-use, LIKWID overwrites their configuration to do its measurements
 - `invalid_to_zero`: In some cases, the calculations result in `NaN` or `Inf`. With this option, all `NaN` and `Inf` values are replaces with `0.0`.
 
@@ -61,7 +63,21 @@ $ scripts/likwid_perfgroup_to_cc_config.py ICX MEM_DP
 }
 ```
 
-You can copy this JSON and add it to the `eventsets` list. If you specify multiple event sets, you can add globally derived metrics in the extra `global_metrics` section with the metric names as variables. 
+You can copy this JSON and add it to the `eventsets` list. If you specify multiple event sets, you can add globally derived metrics in the extra `global_metrics` section with the metric names as variables.
+
+### Mixed usage between daemon and users
+
+LIKWID checks the file `/var/run/likwid.lock` before performing any interfering operations. Who is allowed to access the counters is determined by the owner of the file. If it does not exist, it is created for the current user. So, if you want to temporarly allow counter access to a user (e.g. in a job):
+
+Before (SLURM prolog, ...)
+```
+$ chwon $JOBUSER /var/run/likwid.lock
+```
+
+After (SLURM epilog, ...)
+```
+$ chwon $CCUSER /var/run/likwid.lock
+```
 
 ### Example configuration
 

collectors/lustreMetric.go

@@ -22,6 +22,7 @@ type LustreCollectorConfig struct {
 	LCtlCommand    string   `json:"lctl_command"`
 	ExcludeMetrics []string `json:"exclude_metrics"`
 	SendAllMetrics bool     `json:"send_all_metrics"`
+	Sudo           bool     `json:"use_sudo"`
 }
 
 type LustreCollector struct {
@@ -31,11 +32,17 @@ type LustreCollector struct {
 	stats   map[string]map[string]int64
 	config  LustreCollectorConfig
 	lctl    string
+	sudoCmd string
 }
 
 func (m *LustreCollector) getDeviceDataCommand(device string) []string {
+	var command *exec.Cmd
 	statsfile := fmt.Sprintf("llite.%s.stats", device)
-	command := exec.Command(m.lctl, LCTL_OPTION, statsfile)
+	if m.config.Sudo {
+		command = exec.Command(m.sudoCmd, m.lctl, LCTL_OPTION, statsfile)
+	} else {
+		command = exec.Command(m.lctl, LCTL_OPTION, statsfile)
+	}
 	command.Wait()
 	stdout, _ := command.Output()
 	return strings.Split(string(stdout), "\n")
@@ -136,6 +143,12 @@ func (m *LustreCollector) Init(config json.RawMessage) error {
 		}
 	}
 	m.lctl = p
+	if m.config.Sudo {
+		p, err := exec.LookPath("sudo")
+		if err != nil {
+			m.sudoCmd = p
+		}
+	}
 
 	devices := m.getDevices()
 	if len(devices) == 0 {

scripts/cc-metric-collector.spec

@@ -7,6 +7,7 @@ License:        MIT
 Source0:        %{name}-%{version}.tar.gz
 
 BuildRequires:  go-toolset
+BuildRequires:  systemd-rpm-macros
 # for internal LIKWID installation
 BuildRequires:  wget perl-Data-Dumper
 
@@ -34,11 +35,15 @@ install -Dpm 0600 receivers.json %{buildroot}%{_sysconfdir}/%{name}/receivers.js
 install -Dpm 0600 router.json %{buildroot}%{_sysconfdir}/%{name}/router.json
 install -Dpm 0644 scripts/%{name}.service %{buildroot}%{_unitdir}/%{name}.service
 install -Dpm 0600 scripts/%{name}.config %{buildroot}%{_sysconfdir}/default/%{name}
+install -Dpm 0644 scripts/%{name}.sysusers %{buildroot}%{_sysusersdir}/%{name}.conf
 
 
 %check
 # go test should be here... :)
 
+%pre
+%sysusers_create_compat scripts/%{name}.sysusers
+
 %post
 %systemd_post %{name}.service
 
@@ -55,8 +60,11 @@ install -Dpm 0600 scripts/%{name}.config %{buildroot}%{_sysconfdir}/default/%{na
 %attr(0600,root,root) %config(noreplace) %{_sysconfdir}/%{name}/sinks.json
 %attr(0600,root,root) %config(noreplace) %{_sysconfdir}/%{name}/receivers.json
 %attr(0600,root,root) %config(noreplace) %{_sysconfdir}/%{name}/router.json
+%{_sysusersdir}/%{name}.conf
 
 %changelog
+* Thu Mar 03 2022 Thomas Gruber - 0.3
+- Add clustercockpit user installation
 * Mon Feb 14 2022 Thomas Gruber - 0.2
 - Add component specific configuration files
 - Add %attr to config files

scripts/cc-metric-collector.sysusers

@@ -0,0 +1,2 @@
+#Type Name  ID GECOS                     Home directory           Shell
+s     hpcop -  "User for ClusterCockpit" /run/cc-metric-collector /sbin/nologin