driver: bad connection when no query for hours

[tlightsky]

all our etl service which dong insert into table once per hours will got
this bad connection error
driver version is 1.3.11

[forward32]

CH has idle_connection_timeout option with default value 1 hour. If you do nothing within this time CH closes your connection.

Solutions:

1. just increase this value
2. use separate goroutine with some activity (e.g. ping)

[tlightsky]

@forward32
we suspect that too,
and we tried a db.Ping() every 30 second,
still got the bad connection error,
maybe we need to select now() rather than db.Ping()

[forward32]

Hm, I use second approach (ping once in minute) and it works fine for me.

Try to:

1. check your value of idle_connection_timeout
2. check error logs in CH
3. capture traffic between your app and CH and investigate it for some anomalies

[tlightsky]

@forward32
for now, we try to reconnect if it failed for the first time insertion,
will try to investigate the reason later,
thanks for reply

[awskii]

Previously, db.Ping did not return error at all (driver bug), so you needed to SELECT 1; in a for loop, and reconnect, if select fails.
Looks like something never changes, and that's why I'm here.

[kshvakov]

ResetSession added in V2