Hmmm.... intresting. I tried it with two different Ruby libraries (Net::HTTP and RFuzz) with the exact same behaviour. I'll try to see if there's something else I'm missing.
On Fri, Dec 19, 2008 at 12:13 PM, Adam Kocoloski <[email protected]>wrote: > Hi Ruy, I tried to reproduce this using Python's httplib2 client library, > but I find that I can create new databases without CouchDB closing the > connection on me. I just double-checked that the same ephemeral port was > used for the whole sequence of requests. I did this using a server running > R12B-5 and the latest CouchDB trunk on localhost. Best, > > Adam > > > On Dec 19, 2008, at 1:48 PM, Ruy Díaz Jara wrote: > > Hi, >> >> I recently posted this question on the user mailing list but I'm thinking >> maybe that wasn't the right place to do it (as I have gotten no answer). >> If >> it is, I apologize for moving it here. >> >> Ruy >> >> ---------- Forwarded message ---------- >> From: Ruy Díaz Jara <[email protected]> >> Date: Thu, Dec 18, 2008 at 10:51 AM >> Subject: Connection reset >> To: [email protected] >> >> >> Hi, >> >> I've recently been working on getting my application to use a connection >> pool to connect to Couch and improve performance by reusing the >> connections >> and avoiding the socket opening/closing cost. Everything seems nice and >> dandy except when I create a new database. Apparently every time I create >> a >> DB, the socket is closed by Couch. Does anyone know if this expected >> behaviour and if so what the rationale behind it is? >> >> With all other operations this does not occur. It's not even the PUT >> action >> (I can put documents in the DB repeatedly reusing the socket). It seems to >> be only when DBs are created. >> >> Thanks >> Ruy >> > >
