这个后来有具体结论吗？
我在K8s 环境也出现了相同的问题
[2022/07/15 15:37:27.802 +08:00] [WARN] [wal.go:712] [“slow fdatasync”] [took=2.290399661s] [expected-duration=1s]
[2022/07/15 15:43:39.389 +08:00] [WARN] [wal.go:712] [“slow fdatasync”] [took=1.159444856s] [expected-duration=1s]
[2022/07/15 15:45:06.054 +08:00] [WARN] [wal.go:712] [“slow fdatasync”] [took=2.498501001s] [expected-duration=1s]
[2022/07/15 15:48:11.562 +08:00] [WARN] [wal.go:712] [“slow fdatasync”] [took=1.901545807s] [expected-duration=1s]
[2022/07/15 16:06:45.109 +08:00] [WARN] [wal.go:712] [“slow fdatasync”] [took=1.078352422s] [expected-duration=1s]
然后 PD 似乎一直在切主， 导致获取TSO时间长度不正常， 大面积需要100ms~500ms, 有些高达1s, 看
日志的意思是写文件慢，不过 我这边使用是ceph