Leif Mortenson wrote:
> Danny Angus wrote:
>
>> Hi all,
>> tailing MySQL logs whilst running James I found a regular "SELECT 1"
>> being
>> run, which I now think is being used to ping the connection.
>> However my question is should it be "pung" (Scottish for "pinged" ;) even
>> when queries are being regularly executed?
>>
>> It also seems to be run only when another query is being run anyway.
>> Could someone explain this?
>>
> The data source validates connections before returning them so that any
> problems with the connection can be discovered before returning the bad
> connection to the user code. This gives the data source the opportunity
> to close the bad connection and create a new good connection and
> transparently return that to the user code so the user code never breaks.
>
> Unfortunately if you write your code so that a connection is allocated
> and then closed for each query, then the number of DB queries is doubled.
>
> I had written a database pool last year which got around this problem by
> keeping track of the last time the connection was released. Then the
> ping was only made if that time was greater than 5 seconds. This made a
> big difference by cutting the pings down to nil when the application is
> heavily loaded. The thinking is that when the queries are spaced close
> together, the fact that the previous query was successful gives the same
> information as the ping would, so it is not necessary.
>
> What about implementing something like this in Avalon's data sources as
> well? Drawbacks?
+1, give me a patch and I can apply it!
--
"They that give up essential liberty to obtain a little temporary safety
deserve neither liberty nor safety."
- Benjamin Franklin
--
To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
