LocalhostBunny/jobs/runner.go
2023-12-01 11:24:22 +01:00

171 lines
5.1 KiB
Go

package jobs
import (
"context"
"fmt"
"github.com/rs/zerolog/log"
"gogs.mikescher.com/BlackForestBytes/goext/exerr"
"gogs.mikescher.com/BlackForestBytes/goext/langext"
"gogs.mikescher.com/BlackForestBytes/goext/mathext"
"gogs.mikescher.com/BlackForestBytes/goext/rfctime"
"gogs.mikescher.com/BlackForestBytes/goext/syncext"
bunny "locbunny"
"locbunny/logic"
"locbunny/models"
"math/rand"
"time"
)
type JobFunction[TData any] func(ctx context.Context, app *logic.Application, lstr *JobListener, data *TData) (int, error)
type JobRunner[TData any] struct {
name string
app *logic.Application
isRunning *syncext.AtomicBool
isStarted bool
interval time.Duration
sigChannel chan string
runTimeout time.Duration
jobFunc JobFunction[TData]
data *TData
}
func NewJobRunner[TData any](app *logic.Application, name string, interval time.Duration, timeout time.Duration, fn JobFunction[TData], data *TData) *JobRunner[TData] {
return &JobRunner[TData]{
app: app,
isRunning: syncext.NewAtomicBool(false),
isStarted: false,
sigChannel: make(chan string),
interval: interval,
runTimeout: timeout,
name: name,
jobFunc: fn,
data: data,
}
}
func (j *JobRunner[TData]) Start() error {
if j.isRunning.Get() {
return exerr.New(bunny.ErrJob, "job already running").Build()
}
if j.isStarted {
return exerr.New(bunny.ErrJob, "job was already started").Build() // re-start after stop is not allowed
}
j.isStarted = true
go j.mainLoop()
return nil
}
func (j *JobRunner[TData]) Stop() {
log.Info().Msg(fmt.Sprintf("Stopping Job [%s]", j.name))
syncext.WriteNonBlocking(j.sigChannel, "stop")
j.isRunning.Wait(false)
log.Info().Msg(fmt.Sprintf("Stopped Job [%s]", j.name))
}
func (j *JobRunner[TData]) Running() bool {
return j.isRunning.Get()
}
func (j *JobRunner[TData]) mainLoop() {
j.isRunning.Set(true)
firstRun := true
for {
interval := j.interval
if firstRun {
// randomize first interval to spread jobs around
perc := mathext.Clamp(rand.Float64(), 0.1, 0.5)
interval = time.Duration(int64(float64(interval) * perc))
}
firstRun = false
signal, okay := syncext.ReadChannelWithTimeout(j.sigChannel, interval)
if okay {
if signal == "stop" {
log.Info().Msg(fmt.Sprintf("Job [%s] received <stop> signal", j.name))
break
} else if signal == "run" {
log.Info().Msg(fmt.Sprintf("Job [%s] received <run> signal", j.name))
continue
} else {
log.Error().Msg(fmt.Sprintf("Received unknown job signal: <%s> in job [%s]", signal, j.name))
}
}
log.Debug().Msg(fmt.Sprintf("Run job [%s]", j.name))
err := j.execute()
if err != nil {
log.Err(err).Msg(fmt.Sprintf("Failed to execute job [%s]: %s", j.name, err.Error()))
}
}
log.Info().Msg(fmt.Sprintf("Job [%s] exiting main-loop", j.name))
j.isRunning.Set(false)
}
func (j *JobRunner[TData]) execute() (err error) {
defer func() {
if rec := recover(); rec != nil {
err = exerr.New(bunny.ErrJob, "Recovered panic in JobRunner::execute").Any("recover", rec).Build()
}
}()
runCtx, cancelRunCtx := context.WithTimeout(context.Background(), j.runTimeout)
defer cancelRunCtx()
jobExec := models.JobExecution{
JobExecutionID: models.NewJobExecutionID(),
JobName: j.name,
StartTime: rfctime.NowRFC3339Nano(),
EndTime: nil,
Changes: 0,
Status: models.JobStatusRunning,
}
lstr := NewJobListener(j.app, jobExec.JobExecutionID, j.name)
lstr.LogInfo("JOB_START", "Job started", nil)
changes, err := langext.RunPanicSafeR2(func() (int, error) { return j.jobFunc(runCtx, j.app, lstr, j.data) })
//goland:noinspection GoTypeAssertionOnErrors
if panicerr, ok := err.(langext.PanicWrappedErr); ok {
jobExec.EndTime = langext.Ptr(rfctime.NowRFC3339Nano())
jobExec.Error = langext.Ptr(panicerr.Error())
jobExec.Status = models.JobStatusFailed
jobExec.Changes = changes
lstr.LogFatal("JOB_PANIC", "Job finished with a panic", langext.H{"msg": panicerr.Error(), "obj": panicerr.ReoveredObj()})
log.Error().Str("panic", panicerr.Error()).Msg(fmt.Sprintf("Job '%s' <%s> finished with panic and %d changes after %f minutes", j.name, jobExec.JobExecutionID, changes, jobExec.Delta().Minutes()))
} else if err != nil {
jobExec.EndTime = langext.Ptr(rfctime.NowRFC3339Nano())
jobExec.Error = langext.Ptr(err.Error())
jobExec.Status = models.JobStatusFailed
jobExec.Changes = changes
lstr.LogFatal("JOB_ERR", "Job finished with an error", langext.H{"msg": err.Error(), "err_obj": exerr.FromError(err).ToAPIJson(false, true, true)})
log.Error().Str("err", err.Error()).Msg(fmt.Sprintf("Job '%s' <%s> finished with an error and %d changes after %f minutes", j.name, jobExec.JobExecutionID, changes, jobExec.Delta().Minutes()))
} else {
jobExec.EndTime = langext.Ptr(rfctime.NowRFC3339Nano())
jobExec.Error = nil
jobExec.Status = models.JobStatusSuccess
jobExec.Changes = changes
lstr.LogInfo("JOB_FINISH", "Job finished", nil)
log.Info().Msg(fmt.Sprintf("Job '%s' <%s> finished successfully with %d changes after %f minutes", j.name, jobExec.JobExecutionID, changes, jobExec.Delta().Minutes()))
}
return nil
}