Reid Chan
This PR is not correct still and more complicated than I thought. Need second thoughts.
Better to provide a new test method or a new UT
so with this patch, even live connections will be cleaned up? isn't the purpose to clean the token expired connections?
> > so with this patch, even live connections will be cleaned up? > > isn't the purpose to clean the token expired connections? > > The connections that are...
not really understand and still confused what's the difference with this block ``` if(forceClean || ((x._2.refCount timeout))) { try{ x._2.connection.close() } catch { case e: IOException => logWarning(s"Fail to close...
>> is updated when the connection is closed hmm, sounds like this logic is wrong, can we just update the timestamp logic here instead of introducing a new para
>> then tsNow - x._2.timestamp > timeout will never be triggered. it is a wrong if your sayings is correct. and in fact what it does it just same to...