VictoriaMetrics/lib/protoparser/prometheus/streamparser.go
Aliaksandr Valialkin c63755c316
lib/writeconcurrencylimiter: improve the logic behind -maxConcurrentInserts limit
Previously the -maxConcurrentInserts was limiting the number of established client connections,
which write data to VictoriaMetrics. Some of these connections could be idle.
Such connections do not consume big amounts of CPU and RAM, so there is a little sense in limiting
the number of such connections. So now the -maxConcurrentInserts command-line option
limits the number of concurrently executed insert requests, not including idle connections.

It is recommended removing -maxConcurrentInserts command-line option, since the default value
for this option should work good for most cases.
2023-01-06 22:20:19 -08:00

208 lines
4.8 KiB
Go

package prometheus
import (
"bufio"
"fmt"
"io"
"sync"
"time"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/protoparser/common"
"github.com/VictoriaMetrics/VictoriaMetrics/lib/writeconcurrencylimiter"
"github.com/VictoriaMetrics/metrics"
)
// ParseStream parses lines with Prometheus exposition format from r and calls callback for the parsed rows.
//
// The callback can be called concurrently multiple times for streamed data from r.
//
// callback shouldn't hold rows after returning.
func ParseStream(r io.Reader, defaultTimestamp int64, isGzipped bool, callback func(rows []Row) error, errLogger func(string)) error {
wcr := writeconcurrencylimiter.GetReader(r)
defer writeconcurrencylimiter.PutReader(wcr)
r = wcr
if isGzipped {
zr, err := common.GetGzipReader(r)
if err != nil {
return fmt.Errorf("cannot read gzipped lines with Prometheus exposition format: %w", err)
}
defer common.PutGzipReader(zr)
r = zr
}
ctx := getStreamContext(r)
defer putStreamContext(ctx)
for ctx.Read() {
uw := getUnmarshalWork()
uw.errLogger = errLogger
uw.ctx = ctx
uw.callback = callback
uw.defaultTimestamp = defaultTimestamp
uw.reqBuf, ctx.reqBuf = ctx.reqBuf, uw.reqBuf
ctx.wg.Add(1)
common.ScheduleUnmarshalWork(uw)
wcr.DecConcurrency()
}
ctx.wg.Wait()
if err := ctx.Error(); err != nil {
return err
}
return ctx.callbackErr
}
func (ctx *streamContext) Read() bool {
readCalls.Inc()
if ctx.err != nil || ctx.hasCallbackError() {
return false
}
ctx.reqBuf, ctx.tailBuf, ctx.err = common.ReadLinesBlock(ctx.br, ctx.reqBuf, ctx.tailBuf)
if ctx.err != nil {
if ctx.err != io.EOF {
readErrors.Inc()
ctx.err = fmt.Errorf("cannot read Prometheus exposition data: %w", ctx.err)
}
return false
}
return true
}
type streamContext struct {
br *bufio.Reader
reqBuf []byte
tailBuf []byte
err error
wg sync.WaitGroup
callbackErrLock sync.Mutex
callbackErr error
}
func (ctx *streamContext) Error() error {
if ctx.err == io.EOF {
return nil
}
return ctx.err
}
func (ctx *streamContext) hasCallbackError() bool {
ctx.callbackErrLock.Lock()
ok := ctx.callbackErr != nil
ctx.callbackErrLock.Unlock()
return ok
}
func (ctx *streamContext) reset() {
ctx.br.Reset(nil)
ctx.reqBuf = ctx.reqBuf[:0]
ctx.tailBuf = ctx.tailBuf[:0]
ctx.err = nil
ctx.callbackErr = nil
}
var (
readCalls = metrics.NewCounter(`vm_protoparser_read_calls_total{type="prometheus"}`)
readErrors = metrics.NewCounter(`vm_protoparser_read_errors_total{type="prometheus"}`)
rowsRead = metrics.NewCounter(`vm_protoparser_rows_read_total{type="prometheus"}`)
)
func getStreamContext(r io.Reader) *streamContext {
select {
case ctx := <-streamContextPoolCh:
ctx.br.Reset(r)
return ctx
default:
if v := streamContextPool.Get(); v != nil {
ctx := v.(*streamContext)
ctx.br.Reset(r)
return ctx
}
return &streamContext{
br: bufio.NewReaderSize(r, 64*1024),
}
}
}
func putStreamContext(ctx *streamContext) {
ctx.reset()
select {
case streamContextPoolCh <- ctx:
default:
streamContextPool.Put(ctx)
}
}
var streamContextPool sync.Pool
var streamContextPoolCh = make(chan *streamContext, cgroup.AvailableCPUs())
type unmarshalWork struct {
rows Rows
ctx *streamContext
callback func(rows []Row) error
errLogger func(string)
defaultTimestamp int64
reqBuf []byte
}
func (uw *unmarshalWork) reset() {
uw.rows.Reset()
uw.ctx = nil
uw.callback = nil
uw.errLogger = nil
uw.defaultTimestamp = 0
uw.reqBuf = uw.reqBuf[:0]
}
func (uw *unmarshalWork) runCallback(rows []Row) {
ctx := uw.ctx
if err := uw.callback(rows); err != nil {
ctx.callbackErrLock.Lock()
if ctx.callbackErr == nil {
ctx.callbackErr = fmt.Errorf("error when processing imported data: %w", err)
}
ctx.callbackErrLock.Unlock()
}
ctx.wg.Done()
}
// Unmarshal implements common.UnmarshalWork
func (uw *unmarshalWork) Unmarshal() {
if uw.errLogger != nil {
uw.rows.UnmarshalWithErrLogger(bytesutil.ToUnsafeString(uw.reqBuf), uw.errLogger)
} else {
uw.rows.Unmarshal(bytesutil.ToUnsafeString(uw.reqBuf))
}
rows := uw.rows.Rows
rowsRead.Add(len(rows))
// Fill missing timestamps with the current timestamp.
defaultTimestamp := uw.defaultTimestamp
if defaultTimestamp <= 0 {
defaultTimestamp = int64(time.Now().UnixNano() / 1e6)
}
for i := range rows {
r := &rows[i]
if r.Timestamp == 0 {
r.Timestamp = defaultTimestamp
}
}
uw.runCallback(rows)
putUnmarshalWork(uw)
}
func getUnmarshalWork() *unmarshalWork {
v := unmarshalWorkPool.Get()
if v == nil {
return &unmarshalWork{}
}
return v.(*unmarshalWork)
}
func putUnmarshalWork(uw *unmarshalWork) {
uw.reset()
unmarshalWorkPool.Put(uw)
}
var unmarshalWorkPool sync.Pool