The only tool you need to know what is happening and how to fix it.
Go alertingexception-trackerissue-trackinglogsmetricsmonitoringobservability-platformopen-telemetryotelsession-replaytelemetrytraces
//go:build telemetry_ch
package chdb
import (
"errors"
"io"
"net"
"strings"
"syscall"
"time"
"github.com/ClickHouse/clickhouse-go/v2"
"github.com/ClickHouse/clickhouse-go/v2/lib/driver"
)
const batchSendMaxAttempts = 3
func SendBatch(query string, fill func(driver.Batch) error) error {
var batch driver.Batch
if err := RetryOnConn(batchSendMaxAttempts, IsConnError, func() error {
b, err := Conn.PrepareBatch(BatchCtx(), query)
batch = b
return err
}); err != nil {
return err
}
if err := fill(batch); err != nil {
batch.Abort()
return err
}
return batch.Send()
}
func RetryOnConn(maxAttempts int, retryable func(error) bool, fn func() error) error {
for attempt := 1; ; attempt++ {
err := fn()
if err == nil || attempt >= maxAttempts || !retryable(err) {
return err
}
time.Sleep(time.Duration(attempt) * 50 * time.Millisecond)
}
}
func IsConnError(err error) bool {
if err == nil {
return false
}
var chErr *clickhouse.Exception
if errors.As(err, &chErr) {
return false
}
if errors.Is(err, io.EOF) || errors.Is(err, io.ErrUnexpectedEOF) ||
errors.Is(err, syscall.ECONNRESET) || errors.Is(err, syscall.EPIPE) ||
errors.Is(err, net.ErrClosed) {
return true
}
msg := strings.ToLower(err.Error())
for _, s := range []string{
"eof",
"broken pipe",
"connection reset",
"reset by peer",
"connection refused",
"no route to host",
"use of closed",
"first block packet",
} {
if strings.Contains(msg, s) {
return true
}
}
return false
}