Constant "transient" errors: `Connection reset by peer`, `The read operation timed out`

[paxcodes]

We are getting these transient errors and they happen constantly.

I know the connection still goes through since our error-reporting system has no database errors. I assume the operation is simply retried. Ideally, we don't encounter these transient errors since they add some latency.

Any insight on what could be causing these transient errors? I have already contacted ClickHouse Cloud support and they don't see any problem in the system and they mention that it could be an issue in the driver.