Gracefully recover from OperationalError in ping (#6)

[nils-borrmann-tacto]

The read_varint() method inside ping() can return an OperationalError when there are problems with the connection. This was not handled inside ping(), which caused a crash (Failed to read data from socket. Likely the connection was closed by the remote. ) instead of gracefully reopening a new connection in a connection pool.

[dmkulazhenko]

Related to #149

[nils-borrmann-tacto]

fyi @dmkulazhenko I've fixed this (and a few other issues) in my fork: https://github.com/nils-borrmann-tacto/asynch (We are using our fork in production and it's been quite stable)

[dmkulazhenko]

fyi @dmkulazhenko I've fixed this (and a few other issues) in my fork: https://github.com/nils-borrmann-tacto/asynch (We are using our fork in production and it's been quite stable)

Thanks @nils-borrmann-tacto, we already planning to do something like that, seems to be asynch really seeks for some active maintainer... I don't know how to say it in right tone, but @long2ice consider selecting some additional maintainer like @nils-borrmann-tacto or @stankudrow, it will really help your project to become more mature and active 🙃

Also I can add, that in some part of services we stick to sqlalchemy + clickhouse, I can recommend this fork https://github.com/nebuly-ai/clickhouse-sqlalchemy 😄