[ci] apt & rpm distribution packages (#97)

* Testing goreleaser * Adding packaging pieces * Testing goreleaser Adding packaging pieces * Adding license bits * Fixing newline * Fixing whitespace * Oops forgot conflict * Adding trace templates to distro package * Fixing spacing * Testing goreleaser and re-adding license * Fixing whitespace * [ci] archive static templates with binary (#95) * archive static templates with binary * update docs * Adding in trace path detection logic * Testing goreleaser Adding packaging pieces Testing goreleaser and re-adding license Fixing whitespace Adding in trace path detection logic Testing goreleaser Adding license bits Oops forgot conflict Adding trace templates to distro package Fixing spacing * trace logic * Update .goreleaser.yaml Fixing bad copy/paste, oops! Co-authored-by: abhinavDhulipala <[email protected]> * Removing systemd files for dist packages * Update .goreleaser.yaml Fixing bad copy/paste, oops! Co-authored-by: abhinavDhulipala <[email protected]> make systemctl edit copy/pastable --------- Co-authored-by: abhinavDhulipala <[email protected]>
rivosinc · Sep 10, 2024 · e285f3e · e285f3e
1 parent 3afae44
commit e285f3e
Show file tree

Hide file tree

Showing 4 changed files with 162 additions and 19 deletions.
diff --git a/.goreleaser.yaml b/.goreleaser.yaml
@@ -1,7 +1,6 @@
 # SPDX-FileCopyrightText: 2023 Rivos Inc.
 #
 # SPDX-License-Identifier: Apache-2.0
-
 version: 2
 builds:
   - goos:
@@ -10,7 +9,22 @@ builds:
       - amd64
       - arm
       - arm64
-
+nfpms:
+  - vendor: rivosinc
+    maintainer: abhinavDhulipala
+    contents:
+      - src: exporter/templates
+        dst: /usr/share/prometheus-slurm-exporter/templates
+        file_info:
+          mode: 0644
+          group: root
+          owner: root
+    formats:
+      - apk
+      - deb
+      - rpm
+      - termux.deb
+      - archlinux
 archives:
   - name_template: '{{ .ProjectName }}_{{ .Os }}_{{ .Arch }}'
     files:

diff --git a/README.md b/README.md
@@ -154,6 +154,96 @@ Env vars can be sepcified in a `.env` file, while using the `just`
 | CLI_TIMEOUT     | 10.           | # seconds before the exporter terminates command.                           |
 | TRACE_ROOT_PATH | "cwd"         | path to ./templates directory where html files are located                  |
 
+### RPM/DEB Packages
+
+You can download RPM or DEB versions from the [Releases](releases/) tab. The
+packages are configured to use systemd to start and stop the service.
+
+Configuring the systemd service
+
+`$ systemctl edit prometheus-slurm-exporter.service`
+
+```text
+### Editing /etc/systemd/system/prometheus-slurm-exporter.service.d/override.conf
+### Anything between here and the comment below will become the new contents of the file
+
+[Service]
+Environment="PATH=/opt/slurm/bin"
+Environment="POLL_INTERVAL=300"
+Environment="CLI_TIMEOUT=60"
+Environment="LOGLEVEL=debug"
+
+### Lines below this comment will be discarded
+
+### /usr/lib/systemd/system/prometheus-slurm-exporter.service
+# [Unit]
+# Description=Prometheus SLURM Exporter
+#
+# [Service]
+# ExecStart=/usr/bin/prometheus-slurm-exporter
+# Restart=always
+# RestartSec=15
+#
+# [Install]
+# WantedBy=multi-user.target
+```
+
+### RPM/DEB Packages
+
+You can download RPM or DEB versions from the [Releases](releases/) tab. The
+packages are configured to use systemd to start and stop the service.
+
+### Running with Systemd
+
+You can install a systemd service definition using the following command:
+
+```bash
+sudo bash -c 'cat << EOF > /etc/systemd/system/prometheus-slurm-exporter.service
+[Unit]
+Description=Prometheus SLURM Exporter
+
+[Service]
+ExecStart=/usr/bin/prometheus-slurm-exporter
+Restart=always
+RestartSec=15
+
+[Install]
+WantedBy=multi-user.target
+EOF'
+sudo systemctl daemon-reload
+sudo systemctl enable --now prometheus-slurm-exporter.service
+```
+
+Customizing the systemd service with environment variables:
+
+```bash
+sudo systemctl edit prometheus-slurm-exporter.service`
+```
+
+```text
+### Editing /etc/systemd/system/prometheus-slurm-exporter.service.d/override.conf
+### Anything between here and the comment below will become the new contents of the file
+
+[Service]
+Environment="PATH=/opt/slurm/bin"
+Environment="POLL_INTERVAL=300"
+Environment="CLI_TIMEOUT=60"
+Environment="LOGLEVEL=debug"
+
+### Lines below this comment will be discarded
+
+### /usr/lib/systemd/system/prometheus-slurm-exporter.service
+# [Unit]
+# Description=Prometheus SLURM Exporter
+#
+# [Service]
+# ExecStart=/usr/bin/prometheus-slurm-exporter
+# Restart=always
+# RestartSec=15
+#
+# [Install]
+# WantedBy=multi-user.target
+```
 
 ### Future work
 slurmrestd support
diff --git a/exporter/trace.go b/exporter/trace.go
@@ -8,7 +8,6 @@ import (
 	"encoding/json"
 	"errors"
 	"fmt"
-	"io/fs"
 	"log"
 	"net/http"
 	"os"
@@ -22,8 +21,10 @@ import (
 )
 
 // cleanup on add if greater than this threshold
-const cleanupThreshold uint64 = 1_000
-const templateDirName string = "templates"
+const (
+	cleanupThreshold uint64 = 1_000
+	templateDirName  string = "templates"
+)
 
 // store a jobs published proc stats
 type TraceInfo struct {
@@ -107,22 +108,11 @@ type TraceCollector struct {
 
 func NewTraceCollector(config *Config) *TraceCollector {
 	traceConfig := config.TraceConf
-	templateRootDir := "."
-	// path to look for the /templates directory. Defaults to cwd
-	if path, ok := os.LookupEnv("TRACE_ROOT_PATH"); ok {
-		templateRootDir = path
-	}
-	traceDir := ""
-	err := filepath.WalkDir(templateRootDir, func(path string, d fs.DirEntry, err error) error {
-		if err == nil && d.IsDir() && d.Name() == templateDirName {
-			traceDir = path
-			return nil
-		}
-		return nil
-	})
-	if err != nil || traceDir == "" {
+	traceDir := detectTraceTemplatePath()
+	if traceDir == "" {
 		log.Fatal("no template found")
 	}
+	slog.Debug("using trace template path: " + traceDir)
 	return &TraceCollector{
 		ProcessFetcher: NewAtomicProFetcher(traceConfig.rate),
 		squeueFetcher:  traceConfig.sharedFetcher,
@@ -200,3 +190,24 @@ func (c *TraceCollector) uploadTrace(w http.ResponseWriter, r *http.Request) {
 		}
 	}
 }
+
+// detectTraceTemplatePath returns the trace_root path based on the following criteria:
+// 1. If TRACE_ROOT_PATH is specified, search that directory. If we don't find a templates dir, let's panic and crash the program.
+// 2. If TRACE_ROOT_PATH isn't specified, we can search cwd and /usr/share/prometheus-slurm-exporter.
+// If no templates path is found, returns an empty string
+func detectTraceTemplatePath() string {
+	if rpath, ok := os.LookupEnv("TRACE_ROOT_PATH"); ok {
+		templateP := filepath.Join(rpath, templateDirName)
+		if _, err := os.Stat(templateP); err != nil {
+			panic("TRACE_ROOT_PATH must include a directory called: templates")
+		}
+		return templateP
+	}
+	for _, rpath := range []string{".", "/usr/share/prometheus-slurm-exporter"} {
+		templateP := filepath.Join(rpath, templateDirName)
+		if _, err := os.Stat(templateP); err == nil {
+			return templateP
+		}
+	}
+	return ""
+}
diff --git a/exporter/trace_test.go b/exporter/trace_test.go
@@ -10,11 +10,13 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"os"
+	"path/filepath"
 	"testing"
 	"time"
 
 	"github.com/prometheus/client_golang/prometheus"
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestAtomicFetcher_Cleanup(t *testing.T) {
@@ -26,6 +28,7 @@ func TestAtomicFetcher_Cleanup(t *testing.T) {
 	fetcher.cleanup()
 	assert.Contains(fetcher.Info, int64(10))
 }
+
 func TestAtomicFetcher_Add(t *testing.T) {
 	assert := assert.New(t)
 	fetcher := NewAtomicProFetcher(10)
@@ -202,3 +205,28 @@ func TestPython3Wrapper(t *testing.T) {
 	json.Unmarshal(wrapperOut, &info)
 	assert.Equal(int64(10), info.JobId)
 }
+
+func TestDetectTraceRootPath_Env(t *testing.T) {
+	os.Clearenv()
+	testDir := t.TempDir()
+	t.Setenv("TRACE_ROOT_PATH", testDir)
+	// Ensure that the function panics if given a TRACE_ROOT_PATh with no 'templates' subdirectory
+	assert.PanicsWithValue(t, "TRACE_ROOT_PATH must include a directory called: templates", func() { detectTraceTemplatePath() })
+	require.NoError(t, os.Mkdir(filepath.Join(testDir, templateDirName), 0o700))
+
+	// Now that we have a 'templates' subdir, it should no longer panic
+	assert.Equal(t, filepath.Join(testDir, templateDirName), detectTraceTemplatePath())
+}
+
+func TestDetectTraceRootPath_Default(t *testing.T) {
+	os.Clearenv()
+	testDir := t.TempDir()
+	os.Chdir(testDir)
+
+	// Should come back empty if since we don't yet have a 'templates' subdir
+	assert.Equal(t, detectTraceTemplatePath(), "")
+	require.NoError(t, os.Mkdir(filepath.Join(testDir, templateDirName), 0o700))
+
+	// Now that we have 'templates' subdir, cwd is a valid path
+	assert.Equal(t, templateDirName, detectTraceTemplatePath())
+}