Incorrect uptime in grafana when tiflash meets supersize tables

[Lloyd-Pottiger]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

create a lot of tables in a TiDB cluster (maybe more than 100k).

2. What did you expect to see? (Required)

uptime in grafana panel is correct.

3. What did you see instead (Required)

uptime is incorrect, keep as 8 minutes as time goes by.

4. What is your TiFlash version? (Required)

v6.1.0

[Lloyd-Pottiger]

maybe due to the bug of linux kernel which cause monotonic clock may not be monotonic sometimes. It may cause some operations to fail with "Timeout exceeded: elapsed 18446744073.709553 seconds".Takes previously returned value and returns it again if time stepped back for some reason.
ref to:
https://github.com/ClickHouse/ClickHouse/blob/master/src/Common/Stopwatch.h#L19-L21