How to implement mysql proxy?

This is not a feature request, just a question. I have sphinxsearch with sql interface, and need to do transparent search index rebuild on configs change. Idea is to create second daemon for pushing new data, but use old one for search request until reindexing complete. Then reroute all “read” requests to new daemon.

As far as i understand, it would be convenient to have proxy, that can properly route sql requests and hide details from application.

Is it possible to do something like this with mysql2? I mean, it would be nice to serialize sql requests to some intermediate format (~ like AST in DOM) instead of binary protocol. Or can such task (proxy with intelligent routing) be implemented with different approach?

[puzrin]

Thanks! We already migrated to mysql2 in forum convertor. General impression are:

• seems to work without problems (5M posts conversion).
• returning array container in promise interface looks a bit unusual, but i understand your reasons and that’s not critical in my case.

[sidorares]

@puzrin yeah if you expect this might happen ( connection does not respond for a very long time but TCP socket is alive ) then 1) connection commands queue start growing if you use connection directly or 2) pool callback queue would start growing if limit is not set

It’s still a useful feature, I’m keen to add it, but meanwhile you can have something like this:

   pool.getConnection( (err, conn) => {
      var timeout = setTimeout( () => {
          conn.close(); // this should trigger  removeFromPool - https://github.com/sidorares/node-mysql2/blob/b2010f99180d8b49eb7e9c2b6aeb8cb441362835/lib/pool_connection.js#L17-L21
      }, 30000);

      conn.query(sql, (err, res) => {
         conn.release();
         clearTimeout(timeout);
         // use result;
      });
   });