mirror of
https://github.com/akvorado/akvorado.git
synced 2025-12-12 06:24:10 +01:00
ac68c5970e
This change split the inlet component into a simpler inlet and a new outlet component. The new inlet component receive flows and put them in Kafka, unparsed. The outlet component takes them from Kafka and resume the processing from here (flow parsing, enrichment) and puts them in ClickHouse. The main goal is to ensure the inlet does a minimal work to not be late when processing packets (and restart faster). It also brings some simplification as the number of knobs to tune everything is reduced: for inlet, we only need to tune the queue size for UDP, the number of workers and a few Kafka parameters; for outlet, we need to tune a few Kafka parameters, the number of workers and a few ClickHouse parameters. The outlet component features a simple Kafka input component. The core component becomes just a callback function. There is also a new ClickHouse component to push data to ClickHouse using the low-level ch-go library with batch inserts. This processing has an impact on the internal representation of a FlowMessage. Previously, it was tailored to dynamically build the protobuf message to be put in Kafka. Now, it builds the batch request to be sent to ClickHouse. This makes the FlowMessage structure hides the content of the next batch request and therefore, it should be reused. This also changes the way we decode flows as they don't output FlowMessage anymore, they reuse one that is provided to each worker. The ClickHouse tables are slightly updated. Instead of using Kafka engine, the Null engine is used instead. Fix #1122
111 lines
2.6 KiB
Go
111 lines
2.6 KiB
Go
// SPDX-FileCopyrightText: 2025 Free Mobile
|
|
// SPDX-License-Identifier: AGPL-3.0-only
|
|
|
|
package schema
|
|
|
|
import (
|
|
"net/netip"
|
|
|
|
"github.com/ClickHouse/ch-go/proto"
|
|
"github.com/bits-and-blooms/bitset"
|
|
)
|
|
|
|
// FlowMessage is the abstract representation of a flow through various subsystems.
|
|
type FlowMessage struct {
|
|
TimeReceived uint32
|
|
SamplingRate uint64
|
|
|
|
// For exporter classifier
|
|
ExporterAddress netip.Addr
|
|
|
|
// For interface classifier
|
|
InIf uint32
|
|
OutIf uint32
|
|
SrcVlan uint16
|
|
DstVlan uint16
|
|
|
|
// For routing component
|
|
SrcAddr netip.Addr
|
|
DstAddr netip.Addr
|
|
NextHop netip.Addr
|
|
|
|
// Core component may override them
|
|
SrcAS uint32
|
|
DstAS uint32
|
|
SrcNetMask uint8
|
|
DstNetMask uint8
|
|
|
|
// Only for tests
|
|
OtherColumns map[ColumnKey]any
|
|
|
|
batch clickhouseBatch
|
|
schema *Schema
|
|
}
|
|
|
|
// clickhouseBatch stores columns for efficient streaming. It is embedded
|
|
// inside a FlowMessage.
|
|
type clickhouseBatch struct {
|
|
columns []proto.Column // Indexed by ColumnKey
|
|
columnSet bitset.BitSet // Track which columns have been set
|
|
rowCount int // Number of rows accumulated
|
|
input proto.Input // Input including all columns to stream to ClickHouse
|
|
}
|
|
|
|
// reset resets a flow message. All public fields are set to 0,
|
|
// but the current ClickHouse batch is left untouched.
|
|
func (bf *FlowMessage) reset() {
|
|
*bf = FlowMessage{
|
|
batch: bf.batch,
|
|
schema: bf.schema,
|
|
}
|
|
bf.batch.columnSet.ClearAll()
|
|
}
|
|
|
|
// Clear clears all column data.
|
|
func (bf *FlowMessage) Clear() {
|
|
bf.reset()
|
|
bf.batch.input.Reset()
|
|
bf.batch.rowCount = 0
|
|
}
|
|
|
|
// ClickHouseProtoInput returns the proto.Input that can be used to stream results
|
|
// to ClickHouse.
|
|
func (bf *FlowMessage) ClickHouseProtoInput() proto.Input {
|
|
return bf.batch.input
|
|
}
|
|
|
|
// NewFlowMessage creates a new FlowMessage for the given schema with ClickHouse batch initialized.
|
|
func (schema *Schema) NewFlowMessage() *FlowMessage {
|
|
bf := &FlowMessage{
|
|
schema: schema,
|
|
}
|
|
|
|
maxKey := ColumnKey(0)
|
|
for _, column := range bf.schema.columns {
|
|
if column.Key > maxKey {
|
|
maxKey = column.Key
|
|
}
|
|
}
|
|
|
|
bf.batch.columns = make([]proto.Column, maxKey+1)
|
|
bf.batch.columnSet = *bitset.New(uint(maxKey + 1))
|
|
bf.batch.rowCount = 0
|
|
|
|
for _, column := range bf.schema.columns {
|
|
if !column.Disabled && column.shouldProvideValue() {
|
|
bf.batch.columns[column.Key] = column.newProtoColumn()
|
|
bf.batch.input = append(bf.batch.input, proto.InputColumn{
|
|
Name: column.Name,
|
|
Data: column.wrapProtoColumn(bf.batch.columns[column.Key]),
|
|
})
|
|
}
|
|
}
|
|
|
|
return bf
|
|
}
|
|
|
|
// FlowCount return the number of flows batched
|
|
func (bf *FlowMessage) FlowCount() int {
|
|
return bf.batch.rowCount
|
|
}
|