non-blocking IO in clients (and other questions)

Brian Aker brian at
Tue Oct 2 22:17:23 UTC 2007


On Oct 2, 2007, at 10:59 AM, Dustin Sallings wrote:

>> 2)  The protocol always sends an ACK of some sort.
> 	The interface provided to my client doesn't require the caller to  
> wait for ACKs.  You tend to want to do that for get requests, but  
> you may not care in the case of deletes or sets.

I was thinking about doing exactly this. I pulled the source to the  
perl client this morning to figure out how it works after I got some  
feedback that the Ruby client is currently faster then mine. The perl  
code sets non-blocking, but then loops for the read to handle the  
data (since it always responds with an answer (and this was confirmed  
by Jamie McCarthy). Somewhat defeating the purpose of non-blocking  
unless you open up multiple sockets.

> 	That is to say, you generally don't want to not know when  
> something is over (in the case of quiet gets in the binary  
> protocol, you'll want a noop or a regular get at the end), but you  
> can't really send a quiet get and then wait just in case something  
> starts arriving.  Instead, just stream requests out and stream  
> responses in.  Line them up, and

Yes, I already do this for gets currently. The nature of the protocol  
makes this very natural.

>> 	You don't have to at all.  A set is issued, and the state of the  
>> op is changed to waiting_for_response or something and it's added  
>> to an input queue.  Then you start sending the next operation from  
>> your output queue.  If a server starts sending stuff back to you,  
>> it's for whatever's on the top of your input queue (in the binary  
>> protocol, you can double-check this).

I was thinking to myself this morning that I was hoping the binary  
protocol would be smart enough to do this.

So I see what you are doing, you have created an API which doesn't  
provide for a user to set/get an answer unless they want to. It is  
easy enough to add that sort of call but I am left wandering what  
users expect. Giving them both options is fine. The perl driver makes  
me think that users do expect a "this is how it is going to be". And  
from looking at it, they expect a send/ack.

A lot of what I am looking at if I have a question is the Perl driver  
at the moment. I've got a copy of the Ruby one as well, but from  
looking at its code, it looks to be blocking as well.

>> On a different related note, I've noticed another issue with  
>> "set". When I send a "set foo 0 0 20\r\n", I have to just send  
>> that message. I can't just drop the "set" and the data to be  
>> stored in the same socket. If I do that, then the server removes  
>> whatever portion of the key that was contained in the "set". Maybe  
>> this is my bug (though I can demonstrate it), but that seems like  
>> a waste. AKA if on the server its doing a read() for the set and  
>> tossing out the rest of the packet then its purposely causing two  
>> roundtrips for the same data.
> 	By ``socket,'' do you mean ``packet?''  My client pipelines  
> request in such a way that multiple gets, sets, deletes, etc... can  
> easily get stuffed into the same packet.

Interesting. Do you have a test case of sending the SET and the value  
from within the same write()/send() call?

> 	We can create a qset, but the semantics would need to be carefully  
> considered.  qget just keeps its errors silent and only returns  
> positive results.  Should a qset do the opposite, or should it  
> never return anything at all?

Off the top of my head what we would want is a way to send data, and  
then receive back asynchronously what committed and an "end of  
transaction" statement. From that we can deduce what was what.

Thanks Dustin!


Brian "Krow" Aker, brian at
Seattle, Washington                     <-- Me                <-- Software    <-- Fun
You can't grep a dead tree.

-------------- next part --------------
An HTML attachment was scrubbed...

More information about the memcached mailing list