feat: system monitoring + push agent (merge plan M3)

systems + system_stats tables; agents push JSON every N seconds
to POST /api/systems/ingest with a per-system bearer token. The
server stores each sample, keeps a `latest` snapshot for cheap
list/widget reads, prunes stats after 7d, and marks systems down
when silent > max(3x interval, 30s) — firing system.up/down
webhook events on transitions.

cmd/dash-agent is a standalone Linux collector: /proc for
cpu/mem/net/load/uptime, /sys hwmon temps, statfs disk, and
container liveness via /var/run/docker.sock when present.
No imports from the server module — the sample shape is
duplicated deliberately so the agent stays a single static
binary.

Frontend: /systems card grid with usage bars, /systems/:id with
recharts cpu/mem/net/load, temps + containers tables, and the
agent command; `system` board widget; nav + palette + WidgetDialog
picker wired.

Generated with [Devin](https://devin.ai)

Co-Authored-By: Devin <158243242+devin-ai-integration[bot]@users.noreply.github.com>
This commit is contained in:
Tomas DvorakandDevin committed 2026-09-21 19:02:58 +02:00
1 parent 2f3e8a1879
commit 74c987c18a
16 files changed
+1541 -10

No files matched your search

+9
View File
@@ -41,6 +41,7 @@ func newServer(logger *zap.Logger, db *sql.DB, iconsDir string) *Server {
s.lookup = domain.NewLookup().Run
widget.Register(&monitorWidget{db: db})
widget.Register(&domainWidget{db: db})
widget.Register(&systemWidget{db: db})
return s
}
@@ -92,6 +93,14 @@ func (s *Server) routes() *gin.Engine {
v1.GET("/domains/:id/checks", s.domainChecks)
v1.POST("/notify/test", s.notifyTest)
v1.GET("/systems", s.listSystemsH)
v1.POST("/systems", s.createSystem)
v1.GET("/systems/:id", s.getSystem)
v1.PATCH("/systems/:id", s.patchSystem)
v1.DELETE("/systems/:id", s.deleteSystem)
v1.GET("/systems/:id/stats", s.systemStats)
v1.POST("/systems/ingest", s.ingestSystem)
v1.GET("/status", s.getStatus)
v1.GET("/widgets/types", s.widgetTypes)
v1.GET("/widgets/:id/data", s.widgetData)
+96
View File
@@ -539,6 +539,102 @@ func TestNotifyWebhook(t *testing.T) {
}
}
func TestSystemIngest(t *testing.T) {
srv := testServer(t)
defer srv.Close()
// No token -> 401.
res, _ := do(t, "POST", srv.URL+"/api/systems/ingest", `{"cpu":1}`)
if res.StatusCode != http.StatusUnauthorized {
t.Fatalf("expected 401, got %d", res.StatusCode)
}
res, sys := do(t, "POST", srv.URL+"/api/systems", `{"name":"nas"}`)
if res.StatusCode != http.StatusCreated {
t.Fatalf("create: %v", sys)
}
id, token := sys["id"].(string), sys["token"].(string)
if token == "" {
t.Fatal("no agent token issued")
}
// Bad token -> 401.
req, _ := http.NewRequest("POST", srv.URL+"/api/systems/ingest",
strings.NewReader(`{"cpu":50}`))
req.Header.Set("Authorization", "Bearer wrong")
if r, err := http.DefaultClient.Do(req); err != nil || r.StatusCode != http.StatusUnauthorized {
t.Fatalf("bad token: %v %v", r.StatusCode, err)
}
// Real push -> system goes up, latest + stats stored.
push := func() (*http.Response, map[string]any) {
t.Helper()
req, _ := http.NewRequest("POST", srv.URL+"/api/systems/ingest",
strings.NewReader(`{"host":"nas.local","os":"linux","arch":"amd64",
"cpuModel":"Test CPU","cores":4,"cpu":42.5,"memTotal":8000,"memUsed":4000,
"diskTotal":100000,"diskUsed":50000,"netRx":100,"netTx":50,"load1":1.5}`))
req.Header.Set("Content-Type", "application/json")
req.Header.Set("Authorization", "Bearer "+token)
res, err := http.DefaultClient.Do(req)
if err != nil {
t.Fatalf("push: %v", err)
}
defer res.Body.Close()
var out map[string]any
_ = json.NewDecoder(res.Body).Decode(&out)
return res, out
}
if res, out := push(); res.StatusCode != http.StatusOK {
t.Fatalf("ingest: %v", out)
}
res, sys = do(t, "GET", srv.URL+"/api/systems/"+id, "")
if res.StatusCode != http.StatusOK || sys["status"] != "up" {
t.Fatalf("get: %v", sys)
}
if sys["host"] != "nas.local" || sys["cores"].(float64) != 4 {
t.Fatalf("host fields not stored: %v", sys)
}
latest := sys["latest"].(map[string]any)
if latest["cpu"].(float64) != 42.5 {
t.Fatalf("latest not stored: %v", latest)
}
res, list := do(t, "GET", srv.URL+"/api/systems/"+id+"/stats?hours=1", "")
if res.StatusCode != http.StatusOK {
t.Fatalf("stats: %v", list)
}
// stats returns an array — decode separately.
req2, _ := http.NewRequest("GET", srv.URL+"/api/systems/"+id+"/stats?hours=1", nil)
r2, _ := http.DefaultClient.Do(req2)
var arr []map[string]any
_ = json.NewDecoder(r2.Body).Decode(&arr)
r2.Body.Close()
if len(arr) != 1 || arr[0]["cpu"].(float64) != 42.5 {
t.Fatalf("stats rows: %v", arr)
}
// Widget fetcher serves the latest sample.
_, a := do(t, "POST", srv.URL+"/api/sections", `{"name":"A"}`)
_, w := do(t, "POST", srv.URL+"/api/items", `{"sectionId":"`+a["id"].(string)+`","kind":"widget","name":"S","config":{"type":"system","systemId":"`+id+`"}}`)
res, data := do(t, "GET", srv.URL+"/api/widgets/"+w["id"].(string)+"/data", "")
inner, _ := data["data"].(map[string]any)
if res.StatusCode != http.StatusOK || inner["name"] != "nas" || inner["cpu"].(float64) != 42.5 {
t.Fatalf("system widget: %d %v", res.StatusCode, data)
}
// Rename + token rotation.
res, sys = do(t, "PATCH", srv.URL+"/api/systems/"+id, `{"name":"nas2","rotateToken":true}`)
if sys["name"] != "nas2" || sys["token"] == token {
t.Fatalf("patch: %v", sys)
}
res, _ = do(t, "DELETE", srv.URL+"/api/systems/"+id, "")
if res.StatusCode != http.StatusNoContent {
t.Fatalf("delete: %d", res.StatusCode)
}
}
func TestMidpoint(t *testing.T) {
f := func(v float64) *float64 { return &v }
if got := midpoint(nil, nil); got != 1024 {
+5
View File
@@ -411,6 +411,10 @@ func (s *Server) runScheduler(done <-chan struct{}) {
if _, err := s.db.Exec(`DELETE FROM heartbeats WHERE checked_at < datetime('now','-30 days')`); err != nil {
s.log.Error("heartbeat prune", zap.Error(err))
}
// System samples are high-rate; a week of raw ticks is enough.
if _, err := s.db.Exec(`DELETE FROM system_stats WHERE ts < datetime('now','-7 days')`); err != nil {
s.log.Error("system stats prune", zap.Error(err))
}
}
}
}
@@ -453,6 +457,7 @@ func (s *Server) sweep() {
}
s.sweepDomains()
s.sweepSystems()
}
// runCheck executes one monitor check (with retries) and persists the result.
+343
View File
@@ -0,0 +1,343 @@
package api
import (
"context"
"database/sql"
"encoding/json"
"errors"
"fmt"
"net/http"
"strconv"
"strings"
"github.com/gin-gonic/gin"
"github.com/tdvorak/dash/internal/widget"
"go.uber.org/zap"
)
// ---------- model ----------
// StatSample is one agent payload. Numeric fields are optional-ish: the
// agent may omit what a platform cannot provide.
type StatSample struct {
Host string `json:"host"`
OS string `json:"os"`
Arch string `json:"arch"`
CPUModel string `json:"cpuModel"`
Cores int `json:"cores"`
IntervalS int `json:"intervalS"`
UptimeS float64 `json:"uptimeS"`
CPU float64 `json:"cpu"` // percent
MemTotal float64 `json:"memTotal"` // bytes
MemUsed float64 `json:"memUsed"`
SwapTotal float64 `json:"swapTotal"`
SwapUsed float64 `json:"swapUsed"`
DiskTotal float64 `json:"diskTotal"` // bytes, root fs
DiskUsed float64 `json:"diskUsed"`
NetRx float64 `json:"netRx"` // bytes/sec
NetTx float64 `json:"netTx"`
Load1 float64 `json:"load1"`
Load5 float64 `json:"load5"`
Load15 float64 `json:"load15"`
Temps map[string]float64 `json:"temps"`
Containers []ContainerStat `json:"containers"`
}
// ContainerStat is one Docker container's liveness; per-container CPU/mem
// needs cgroup sampling — deferred (v1 ships name+state only).
type ContainerStat struct {
Name string `json:"name"`
State string `json:"state"`
}
// System is one monitored host; `latest` is the newest StatSample.
type System struct {
ID string `json:"id"`
Name string `json:"name"`
Token string `json:"token"`
Host string `json:"host"`
OS string `json:"os"`
Arch string `json:"arch"`
CPUModel string `json:"cpuModel"`
Cores int `json:"cores"`
IntervalS int `json:"intervalS"`
Status string `json:"status"` // pending | up | down
LastSeenAt *string `json:"lastSeenAt"`
Latest *StatSample `json:"latest"`
CreatedAt string `json:"createdAt"`
}
const systemCols = `id, name, token, host, os, arch, cpu_model, cores,
interval_s, status, last_seen_at, latest, created_at`
func scanSystem(row interface{ Scan(...any) error }) (*System, error) {
var sys System
var latest string
err := row.Scan(&sys.ID, &sys.Name, &sys.Token, &sys.Host, &sys.OS,
&sys.Arch, &sys.CPUModel, &sys.Cores, &sys.IntervalS, &sys.Status,
&sys.LastSeenAt, &latest, &sys.CreatedAt)
if err != nil {
return nil, err
}
if latest != "" && latest != "{}" {
var s StatSample
if json.Unmarshal([]byte(latest), &s) == nil {
sys.Latest = &s
}
}
return &sys, nil
}
func (s *Server) loadSystem(id string) (*System, error) {
return scanSystem(s.db.QueryRow(`SELECT `+systemCols+` FROM systems WHERE id = ?`, id))
}
func (s *Server) loadSystemByToken(token string) (*System, error) {
return scanSystem(s.db.QueryRow(
`SELECT `+systemCols+` FROM systems WHERE token = ?`, token))
}
// ---------- handlers ----------
func (s *Server) listSystemsH(c *gin.Context) {
rows, err := s.db.Query(`SELECT ` + systemCols + ` FROM systems ORDER BY name`)
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
defer rows.Close()
out := []System{}
for rows.Next() {
sys, err := scanSystem(rows)
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
out = append(out, *sys)
}
c.JSON(http.StatusOK, out)
}
func (s *Server) createSystem(c *gin.Context) {
var in struct {
Name string `json:"name"`
}
if err := c.ShouldBindJSON(&in); err != nil || in.Name == "" {
fail(c, http.StatusBadRequest, "name required")
return
}
id := newID("sys")
if _, err := s.db.Exec(`INSERT INTO systems (id, name, token) VALUES (?,?,?)`,
id, in.Name, newID("agent")); err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
sys, err := s.loadSystem(id)
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
c.JSON(http.StatusCreated, sys)
}
func (s *Server) getSystem(c *gin.Context) {
sys, err := s.loadSystem(c.Param("id"))
if notFound(err) {
fail(c, http.StatusNotFound, "system not found")
return
}
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
c.JSON(http.StatusOK, sys)
}
func (s *Server) patchSystem(c *gin.Context) {
id := c.Param("id")
if _, err := s.loadSystem(id); notFound(err) {
fail(c, http.StatusNotFound, "system not found")
return
}
var in struct {
Name *string `json:"name"`
Rotate *bool `json:"rotateToken"`
}
if err := c.ShouldBindJSON(&in); err != nil {
fail(c, http.StatusBadRequest, "invalid body")
return
}
if in.Name != nil && *in.Name != "" {
_, _ = s.db.Exec(`UPDATE systems SET name = ? WHERE id = ?`, *in.Name, id)
}
if in.Rotate != nil && *in.Rotate {
_, _ = s.db.Exec(`UPDATE systems SET token = ? WHERE id = ?`, newID("agent"), id)
}
sys, err := s.loadSystem(id)
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
c.JSON(http.StatusOK, sys)
}
func (s *Server) deleteSystem(c *gin.Context) {
res, err := s.db.Exec(`DELETE FROM systems WHERE id = ?`, c.Param("id"))
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
if n, _ := res.RowsAffected(); n == 0 {
fail(c, http.StatusNotFound, "system not found")
return
}
c.Status(http.StatusNoContent)
}
// systemStats returns stored samples newest-last for charting.
func (s *Server) systemStats(c *gin.Context) {
if _, err := s.loadSystem(c.Param("id")); notFound(err) {
fail(c, http.StatusNotFound, "system not found")
return
}
hours, _ := strconv.Atoi(c.DefaultQuery("hours", "24"))
if hours <= 0 || hours > 24*7 {
hours = 24
}
rows, err := s.db.Query(`SELECT ts, payload FROM system_stats
WHERE system_id = ? AND ts >= datetime('now', ?) ORDER BY ts`,
c.Param("id"), "-"+strconv.Itoa(hours)+" hours")
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
defer rows.Close()
type sample struct {
TS string `json:"ts"`
StatSample
}
out := []sample{}
for rows.Next() {
var ts, payload string
if err := rows.Scan(&ts, &payload); err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
var sm sample
sm.TS = ts
if json.Unmarshal([]byte(payload), &sm.StatSample) == nil {
out = append(out, sm)
}
}
c.JSON(http.StatusOK, out)
}
// ingestSystem receives one agent push. Auth: Authorization: Bearer <token>.
func (s *Server) ingestSystem(c *gin.Context) {
token := strings.TrimPrefix(c.GetHeader("Authorization"), "Bearer ")
if token == "" {
token = c.GetHeader("X-Dash-Token")
}
if token == "" {
fail(c, http.StatusUnauthorized, "token required")
return
}
sys, err := s.loadSystemByToken(token)
if err != nil {
fail(c, http.StatusUnauthorized, "unknown token")
return
}
var sample StatSample
if err := c.ShouldBindJSON(&sample); err != nil {
fail(c, http.StatusBadRequest, "invalid body")
return
}
if sample.IntervalS <= 0 {
sample.IntervalS = 10
}
payload, _ := json.Marshal(&sample)
prev := sys.Status
_, err = s.db.Exec(`UPDATE systems SET
host = ?, os = ?, arch = ?, cpu_model = ?, cores = ?, interval_s = ?,
status = 'up', last_seen_at = strftime('%Y-%m-%dT%H:%M:%fZ','now'),
latest = ?
WHERE id = ?`,
sample.Host, sample.OS, sample.Arch, sample.CPUModel, sample.Cores,
sample.IntervalS, string(payload), sys.ID)
if err != nil {
fail(c, http.StatusInternalServerError, err.Error())
return
}
if _, err := s.db.Exec(`INSERT INTO system_stats (system_id, payload) VALUES (?,?)`,
sys.ID, string(payload)); err != nil {
s.log.Error("system stats insert", zap.Error(err))
}
if prev != "up" {
s.notify("system.up", sys.Name,
fmt.Sprintf("System %q is back online (%s)", sys.Name, sample.Host))
}
c.JSON(http.StatusOK, gin.H{"status": "ok"})
}
// sweepSystems marks agents down when silent for > 3 intervals (min 30s).
func (s *Server) sweepSystems() {
rows, err := s.db.Query(`UPDATE systems SET status = 'down'
WHERE status = 'up' AND last_seen_at IS NOT NULL
AND datetime(last_seen_at, '+' || MAX(3 * interval_s, 30) || ' seconds') <= datetime('now')
RETURNING id, name, host`)
if err != nil {
s.log.Error("system sweep", zap.Error(err))
return
}
defer rows.Close()
for rows.Next() {
var id, name, host string
if rows.Scan(&id, &name, &host) == nil {
s.notify("system.down", name,
fmt.Sprintf("System %q stopped reporting (%s)", name, host))
}
}
}
// ---------- widget fetcher ----------
// systemWidget serves board tiles bound to a system via config.systemId.
type systemWidget struct{ db *sql.DB }
func (w *systemWidget) Meta() widget.Type {
return widget.Type{
Type: "system", Name: "System",
Description: "Live CPU/memory/disk of a monitored host",
Fields: []widget.Field{
{Key: "systemId", Label: "System", Required: true},
},
}
}
func (w *systemWidget) Fetch(_ context.Context, cfg json.RawMessage) (any, error) {
var c struct {
SystemID string `json:"systemId"`
}
if err := json.Unmarshal(cfg, &c); err != nil || c.SystemID == "" {
return nil, errors.New("config.systemId required")
}
var name, status, latest string
var lastSeen *string
err := w.db.QueryRow(`SELECT name, status, last_seen_at, latest
FROM systems WHERE id = ?`, c.SystemID).Scan(&name, &status, &lastSeen, &latest)
if err != nil {
return nil, errors.New("system not found")
}
var sm StatSample
_ = json.Unmarshal([]byte(latest), &sm)
return map[string]any{
"name": name,
"status": status,
"lastSeen": lastSeen,
"cpu": sm.CPU,
"memTotal": sm.MemTotal,
"memUsed": sm.MemUsed,
"diskTotal": sm.DiskTotal,
"diskUsed": sm.DiskUsed,
}, nil
}