Configure log levels or use custom logger

[Vardominator]

Is it possible to configure the log level or use a custom logger for github.com/tikv/client-go/v2/txnkv? For data-intensive workloads, it is very noisy and difficult to track.

I'm seeing many INFO and WARN logs that are not actionable and clouding useful logs that I actually need to monitor:

...
[INFO] [lock_resolver.go:1081] ["resolveLock rollback"]
...
[INFO] [2pc.go:677] ["[BIG_TXN]"]
...

[aep]

they're also on stdout, making it impossible to build cli utils that use tikv client

[aep]

the logger is a global and wraps zap in a weird inflexible way, but there's a trick:

this is what i do now as a workaround:

import (
	pingcaplog "github.com/pingcap/log"
	"go.uber.org/zap"
)

func init() {
	l, p, _ := pingcaplog.InitLogger(&pingcaplog.Config{})
 
        // p is some sort of wrapper that doesnt do anything but is required for ReplaceGlobals later
        // l is zap, but its messed up, just throw it away and make a new one

	config := zap.NewProductionConfig()
	config.OutputPaths = []string{"stderr"}
	config.ErrorOutputPaths = []string{"stderr"}
	l, _ = config.Build()

	pingcaplog.ReplaceGlobals(l, p)
}