app/vlinsert: add flag to limit amount of fields per line (#4976)

Adding limit on ingestion allows to avoid issues like this one https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4762
Such issues are often caused by misconfigurtion on log persing/ingestion side and preventing such rows from being ingested allows to avoid performance implications created by storing such log rows.

Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>

app/vlinsert/insertutils/common_params.go

@@ -3,12 +3,13 @@ package insertutils
 import (
 	"net/http"
 
+	"github.com/VictoriaMetrics/metrics"
+
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vlstorage"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httputils"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logstorage"
-	"github.com/VictoriaMetrics/metrics"
 )
 
 // CommonParams contains common HTTP parameters used by log ingestion APIs.
@@ -73,12 +74,20 @@ func GetCommonParams(r *http.Request) (*CommonParams, error) {
 // GetProcessLogMessageFunc returns a function, which adds parsed log messages to lr.
 func (cp *CommonParams) GetProcessLogMessageFunc(lr *logstorage.LogRows) func(timestamp int64, fields []logstorage.Field) {
 	return func(timestamp int64, fields []logstorage.Field) {
+		if len(fields) > *MaxFieldsPerLine {
+			var rf logstorage.RowFormatter
+			rf = append(rf[:0], fields...)
+			logger.Warnf("dropping log line with %d fields; it exceeds -insert.maxFieldsPerLine=%d; %s", len(fields), *MaxFieldsPerLine, rf.String())
+			rowsDroppedTotalTooManyFields.Inc()
+			return
+		}
+
 		lr.MustAdd(cp.TenantID, timestamp, fields)
 		if cp.Debug {
 			s := lr.GetRowString(0)
 			lr.ResetKeepSettings()
 			logger.Infof("remoteAddr=%s; requestURI=%s; ignoring log entry because of `debug` query arg: %s", cp.DebugRemoteAddr, cp.DebugRequestURI, s)
-			rowsDroppedTotal.Inc()
+			rowsDroppedTotalDebug.Inc()
 			return
 		}
 		if lr.NeedFlush() {
@@ -88,4 +97,5 @@ func (cp *CommonParams) GetProcessLogMessageFunc(lr *logstorage.LogRows) func(ti
 	}
 }
 
-var rowsDroppedTotal = metrics.NewCounter(`vl_rows_dropped_total{reason="debug"}`)
+var rowsDroppedTotalDebug = metrics.NewCounter(`vl_rows_dropped_total{reason="debug"}`)
+var rowsDroppedTotalTooManyFields = metrics.NewCounter(`vl_rows_dropped_total{reason="too_many_fields"}`)

app/vlinsert/insertutils/flags.go

@@ -1,10 +1,15 @@
 package insertutils
 
 import (
+	"flag"
+
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 )
 
 var (
 	// MaxLineSizeBytes is the maximum length of a single line for /insert/* handlers
 	MaxLineSizeBytes = flagutil.NewBytes("insert.maxLineSizeBytes", 256*1024, "The maximum size of a single line, which can be read by /insert/* handlers")
+
+	// MaxFieldsPerLine is the maximum number of fields per line for /insert/* handlers
+	MaxFieldsPerLine = flag.Int("insert.maxFieldsPerLine", 1000, "The maximum number of fields per line, which can be read by /insert/* handlers")
 )

docs/VictoriaLogs/CHANGELOG.md

@@ -9,6 +9,7 @@ according to [these docs](https://docs.victoriametrics.com/VictoriaLogs/QuickSta
 * FEATURE: expose the following metrics at [/metrics](monitoring) page:
   * `vl_data_size_bytes{type="storage"}` - on-disk size for data excluding [log stream](https://docs.victoriametrics.com/VictoriaLogs/keyConcepts.html#stream-fields) indexes.
   * `vl_data_size_bytes{type="indexdb"}` - on-disk size for [log stream](https://docs.victoriametrics.com/VictoriaLogs/keyConcepts.html#stream-fields) indexes.
+* FEATURE: add `-insert.maxFieldsPerLine` command-line flag, which can be used for limiting the number of fields per line in logs sent to VictoriaLogs via ingestion protocols. This helps to avoid issues like [this](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4762).
 
 * BUGFIX: fix possible panic when no data is written to VictoriaLogs for a long time. See [this issue](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4895). Thanks to @crossoverJie for filing and fixing the issue.
 * BUGFIX: add `/insert/loky/ready` endpoint, which is used by Promtail for healthchecks. This should remove `unsupported path requested: /insert/loki/ready` warning logs. See [this comment](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4762#issuecomment-1690966722).

docs/VictoriaLogs/README.md

@@ -171,6 +171,8 @@ Pass `-help` to VictoriaLogs in order to see the list of supported command-line
     	Whether to use proxy protocol for connections accepted at -httpListenAddr . See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt . With enabled proxy protocol http server cannot serve regular /metrics endpoint. Use -pushmetrics.url for metrics pushing
   -inmemoryDataFlushInterval duration
     	The interval for guaranteed saving of in-memory data to disk. The saved data survives unclean shutdowns such as OOM crash, hardware reset, SIGKILL, etc. Bigger intervals may help increase the lifetime of flash storage with limited write cycles (e.g. Raspberry PI). Smaller intervals increase disk IO load. Minimum supported value is 1s (default 5s)
+  -insert.maxFieldsPerLine int
+    	The maximum number of fields per line, which can be read by /insert/* handlers (default 1000)
   -insert.maxLineSizeBytes size
     	The maximum size of a single line, which can be read by /insert/* handlers
     	Supports the following optional suffixes for size values: KB, MB, GB, TB, KiB, MiB, GiB, TiB (default 262144)