93e4656a27
Before this change workers can only be allocated one GPU task, regardless of how much of the GPU resources that task uses, or how many GPUs are in the system. This makes GPUUtilization a float which can represent that a task needs a portion, or multiple GPUs. GPUs are accounted for like RAM and CPUs so that workers with more GPUs can be allocated more tasks. A known issue is that PC2 cannot use multiple GPUs. And even if the worker has multiple GPUs and is allocated multiple PC2 tasks, those tasks will only run on the first GPU. This could result in unexpected behavior when a worker with multiple GPUs is assigned multiple PC2 tasks. But this should not suprise any existing users who upgrade, as any existing users who run workers with multiple GPUs should already know this and be running a worker per GPU for PC2. But now those users have the freedom to customize the GPU utilization of PC2 to be less than one and effectively run multiple PC2 processes in a single worker. C2 is capable of utilizing multiple GPUs, and now workers can be customized for C2 accordingly.
340 lines
7.9 KiB
Go
340 lines
7.9 KiB
Go
package main
|
|
|
|
import (
|
|
"encoding/hex"
|
|
"encoding/json"
|
|
"fmt"
|
|
"math"
|
|
"os"
|
|
"sort"
|
|
"strings"
|
|
"text/tabwriter"
|
|
"time"
|
|
|
|
"github.com/fatih/color"
|
|
"github.com/google/uuid"
|
|
"github.com/urfave/cli/v2"
|
|
"golang.org/x/xerrors"
|
|
|
|
"github.com/filecoin-project/lotus/extern/sector-storage/storiface"
|
|
|
|
"github.com/filecoin-project/lotus/chain/types"
|
|
lcli "github.com/filecoin-project/lotus/cli"
|
|
)
|
|
|
|
var sealingCmd = &cli.Command{
|
|
Name: "sealing",
|
|
Usage: "interact with sealing pipeline",
|
|
Subcommands: []*cli.Command{
|
|
sealingJobsCmd,
|
|
sealingWorkersCmd,
|
|
sealingSchedDiagCmd,
|
|
sealingAbortCmd,
|
|
},
|
|
}
|
|
|
|
var barCols = float64(64)
|
|
|
|
func barString(total, y, g float64) string {
|
|
yBars := int(math.Round(y / total * barCols))
|
|
gBars := int(math.Round(g / total * barCols))
|
|
eBars := int(barCols) - yBars - gBars
|
|
return color.YellowString(strings.Repeat("|", yBars)) +
|
|
color.GreenString(strings.Repeat("|", gBars)) +
|
|
strings.Repeat(" ", eBars)
|
|
}
|
|
|
|
var sealingWorkersCmd = &cli.Command{
|
|
Name: "workers",
|
|
Usage: "list workers",
|
|
Flags: []cli.Flag{
|
|
&cli.BoolFlag{
|
|
Name: "color",
|
|
Usage: "use color in display output",
|
|
DefaultText: "depends on output being a TTY",
|
|
},
|
|
},
|
|
Action: func(cctx *cli.Context) error {
|
|
if cctx.IsSet("color") {
|
|
color.NoColor = !cctx.Bool("color")
|
|
}
|
|
|
|
nodeApi, closer, err := lcli.GetStorageMinerAPI(cctx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
defer closer()
|
|
|
|
ctx := lcli.ReqContext(cctx)
|
|
|
|
stats, err := nodeApi.WorkerStats(ctx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
type sortableStat struct {
|
|
id uuid.UUID
|
|
storiface.WorkerStats
|
|
}
|
|
|
|
st := make([]sortableStat, 0, len(stats))
|
|
for id, stat := range stats {
|
|
st = append(st, sortableStat{id, stat})
|
|
}
|
|
|
|
sort.Slice(st, func(i, j int) bool {
|
|
return st[i].id.String() < st[j].id.String()
|
|
})
|
|
|
|
for _, stat := range st {
|
|
gpuUse := "not "
|
|
gpuCol := color.FgBlue
|
|
if stat.GpuUsed > 0 {
|
|
gpuCol = color.FgGreen
|
|
gpuUse = ""
|
|
}
|
|
|
|
var disabled string
|
|
if !stat.Enabled {
|
|
disabled = color.RedString(" (disabled)")
|
|
}
|
|
|
|
fmt.Printf("Worker %s, host %s%s\n", stat.id, color.MagentaString(stat.Info.Hostname), disabled)
|
|
|
|
fmt.Printf("\tCPU: [%s] %d/%d core(s) in use\n",
|
|
barString(float64(stat.Info.Resources.CPUs), 0, float64(stat.CpuUse)), stat.CpuUse, stat.Info.Resources.CPUs)
|
|
|
|
ramTotal := stat.Info.Resources.MemPhysical
|
|
ramTasks := stat.MemUsedMin
|
|
ramUsed := stat.Info.Resources.MemUsed
|
|
var ramReserved uint64 = 0
|
|
if ramUsed > ramTasks {
|
|
ramReserved = ramUsed - ramTasks
|
|
}
|
|
ramBar := barString(float64(ramTotal), float64(ramReserved), float64(ramTasks))
|
|
|
|
fmt.Printf("\tRAM: [%s] %d%% %s/%s\n", ramBar,
|
|
(ramTasks+ramReserved)*100/stat.Info.Resources.MemPhysical,
|
|
types.SizeStr(types.NewInt(ramTasks+ramUsed)),
|
|
types.SizeStr(types.NewInt(stat.Info.Resources.MemPhysical)))
|
|
|
|
vmemTotal := stat.Info.Resources.MemPhysical + stat.Info.Resources.MemSwap
|
|
vmemTasks := stat.MemUsedMax
|
|
vmemUsed := stat.Info.Resources.MemUsed + stat.Info.Resources.MemSwapUsed
|
|
var vmemReserved uint64 = 0
|
|
if vmemUsed > vmemTasks {
|
|
vmemReserved = vmemUsed - vmemTasks
|
|
}
|
|
vmemBar := barString(float64(vmemTotal), float64(vmemReserved), float64(vmemTasks))
|
|
|
|
fmt.Printf("\tVMEM: [%s] %d%% %s/%s\n", vmemBar,
|
|
(vmemTasks+vmemReserved)*100/vmemTotal,
|
|
types.SizeStr(types.NewInt(vmemTasks+vmemReserved)),
|
|
types.SizeStr(types.NewInt(vmemTotal)))
|
|
|
|
if len(stat.Info.Resources.GPUs) > 0 {
|
|
gpuBar := barString(float64(len(stat.Info.Resources.GPUs)), 0, stat.GpuUsed)
|
|
fmt.Printf("\tGPU: [%s] %.f%% %.2f/%d gpu(s) in use\n", color.GreenString(gpuBar),
|
|
stat.GpuUsed*100/float64(len(stat.Info.Resources.GPUs)),
|
|
stat.GpuUsed, len(stat.Info.Resources.GPUs))
|
|
}
|
|
for _, gpu := range stat.Info.Resources.GPUs {
|
|
fmt.Printf("\tGPU: %s\n", color.New(gpuCol).Sprintf("%s, %sused", gpu, gpuUse))
|
|
}
|
|
}
|
|
|
|
return nil
|
|
},
|
|
}
|
|
|
|
var sealingJobsCmd = &cli.Command{
|
|
Name: "jobs",
|
|
Usage: "list running jobs",
|
|
Flags: []cli.Flag{
|
|
&cli.BoolFlag{
|
|
Name: "color",
|
|
Usage: "use color in display output",
|
|
DefaultText: "depends on output being a TTY",
|
|
},
|
|
&cli.BoolFlag{
|
|
Name: "show-ret-done",
|
|
Usage: "show returned but not consumed calls",
|
|
},
|
|
},
|
|
Action: func(cctx *cli.Context) error {
|
|
if cctx.IsSet("color") {
|
|
color.NoColor = !cctx.Bool("color")
|
|
}
|
|
|
|
nodeApi, closer, err := lcli.GetStorageMinerAPI(cctx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
defer closer()
|
|
|
|
ctx := lcli.ReqContext(cctx)
|
|
|
|
jobs, err := nodeApi.WorkerJobs(ctx)
|
|
if err != nil {
|
|
return xerrors.Errorf("getting worker jobs: %w", err)
|
|
}
|
|
|
|
type line struct {
|
|
storiface.WorkerJob
|
|
wid uuid.UUID
|
|
}
|
|
|
|
lines := make([]line, 0)
|
|
|
|
for wid, jobs := range jobs {
|
|
for _, job := range jobs {
|
|
lines = append(lines, line{
|
|
WorkerJob: job,
|
|
wid: wid,
|
|
})
|
|
}
|
|
}
|
|
|
|
// oldest first
|
|
sort.Slice(lines, func(i, j int) bool {
|
|
if lines[i].RunWait != lines[j].RunWait {
|
|
return lines[i].RunWait < lines[j].RunWait
|
|
}
|
|
if lines[i].Start.Equal(lines[j].Start) {
|
|
return lines[i].ID.ID.String() < lines[j].ID.ID.String()
|
|
}
|
|
return lines[i].Start.Before(lines[j].Start)
|
|
})
|
|
|
|
workerHostnames := map[uuid.UUID]string{}
|
|
|
|
wst, err := nodeApi.WorkerStats(ctx)
|
|
if err != nil {
|
|
return xerrors.Errorf("getting worker stats: %w", err)
|
|
}
|
|
|
|
for wid, st := range wst {
|
|
workerHostnames[wid] = st.Info.Hostname
|
|
}
|
|
|
|
tw := tabwriter.NewWriter(os.Stdout, 2, 4, 2, ' ', 0)
|
|
_, _ = fmt.Fprintf(tw, "ID\tSector\tWorker\tHostname\tTask\tState\tTime\n")
|
|
|
|
for _, l := range lines {
|
|
state := "running"
|
|
switch {
|
|
case l.RunWait > 1:
|
|
state = fmt.Sprintf("assigned(%d)", l.RunWait-1)
|
|
case l.RunWait == storiface.RWPrepared:
|
|
state = "prepared"
|
|
case l.RunWait == storiface.RWRetDone:
|
|
if !cctx.Bool("show-ret-done") {
|
|
continue
|
|
}
|
|
state = "ret-done"
|
|
case l.RunWait == storiface.RWReturned:
|
|
state = "returned"
|
|
case l.RunWait == storiface.RWRetWait:
|
|
state = "ret-wait"
|
|
}
|
|
dur := "n/a"
|
|
if !l.Start.IsZero() {
|
|
dur = time.Now().Sub(l.Start).Truncate(time.Millisecond * 100).String()
|
|
}
|
|
|
|
hostname, ok := workerHostnames[l.wid]
|
|
if !ok {
|
|
hostname = l.Hostname
|
|
}
|
|
|
|
_, _ = fmt.Fprintf(tw, "%s\t%d\t%s\t%s\t%s\t%s\t%s\n",
|
|
hex.EncodeToString(l.ID.ID[:4]),
|
|
l.Sector.Number,
|
|
hex.EncodeToString(l.wid[:4]),
|
|
hostname,
|
|
l.Task.Short(),
|
|
state,
|
|
dur)
|
|
}
|
|
|
|
return tw.Flush()
|
|
},
|
|
}
|
|
|
|
var sealingSchedDiagCmd = &cli.Command{
|
|
Name: "sched-diag",
|
|
Usage: "Dump internal scheduler state",
|
|
Flags: []cli.Flag{
|
|
&cli.BoolFlag{
|
|
Name: "force-sched",
|
|
},
|
|
},
|
|
Action: func(cctx *cli.Context) error {
|
|
nodeApi, closer, err := lcli.GetStorageMinerAPI(cctx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
defer closer()
|
|
|
|
ctx := lcli.ReqContext(cctx)
|
|
|
|
st, err := nodeApi.SealingSchedDiag(ctx, cctx.Bool("force-sched"))
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
j, err := json.MarshalIndent(&st, "", " ")
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
fmt.Println(string(j))
|
|
|
|
return nil
|
|
},
|
|
}
|
|
|
|
var sealingAbortCmd = &cli.Command{
|
|
Name: "abort",
|
|
Usage: "Abort a running job",
|
|
ArgsUsage: "[callid]",
|
|
Action: func(cctx *cli.Context) error {
|
|
if cctx.Args().Len() != 1 {
|
|
return xerrors.Errorf("expected 1 argument")
|
|
}
|
|
|
|
nodeApi, closer, err := lcli.GetStorageMinerAPI(cctx)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
defer closer()
|
|
|
|
ctx := lcli.ReqContext(cctx)
|
|
|
|
jobs, err := nodeApi.WorkerJobs(ctx)
|
|
if err != nil {
|
|
return xerrors.Errorf("getting worker jobs: %w", err)
|
|
}
|
|
|
|
var job *storiface.WorkerJob
|
|
outer:
|
|
for _, workerJobs := range jobs {
|
|
for _, j := range workerJobs {
|
|
if strings.HasPrefix(j.ID.ID.String(), cctx.Args().First()) {
|
|
j := j
|
|
job = &j
|
|
break outer
|
|
}
|
|
}
|
|
}
|
|
|
|
if job == nil {
|
|
return xerrors.Errorf("job with specified id prefix not found")
|
|
}
|
|
|
|
fmt.Printf("aborting job %s, task %s, sector %d, running on host %s\n", job.ID.String(), job.Task.Short(), job.Sector.Number, job.Hostname)
|
|
|
|
return nodeApi.SealingAbort(ctx, job.ID)
|
|
},
|
|
}
|