W3C home > Mailing lists > Public > ietf-http-wg-old@w3.org > September to December 2001

Re: Server closing socket without notice

From: Jeffrey Mogul <mogul@pa.dec.com>
Date: Wed, 10 Oct 2001 16:09:18 -0700
Message-Id: <200110102309.QAA21531@wera.pa.dec.com>
To: Matt Black <MBlack@Smart421.com>
Cc: "'http-wg@cuckoo.hpl.hp.com'" <http-wg@cuckoo.hpl.hp.com>
Matt Black <MBlack@Smart421.com> writes:

    I am seeing SilverStream servers closing HTTP/1.1 connections
    immediately after a small random number of GET requests without
    notice - no 'connection:  close' response header is returned with
    the last successful request.
    Is this behaviour consistent with
    a) the spirit of the RFC
    b) the letter of the RFC
Answer: (b)

See RFC2616, section 8.1.4 (Practical Considerations):

   A client, server, or proxy MAY close the transport connection at any
   time. For example, a client might have started to send a new request
   at the same time that the server has decided to close the "idle"
   connection. From the server's point of view, the connection is being
   closed while it was idle, but from the client's point of view, a
   request is in progress.

   This means that clients, servers, and proxies MUST be able to recover
   from asynchronous close events.

The use of "Connection: close" is a SHOULD-level requirement in the
case where "the server chooses to close the connection immediately
after sending the response" (section, but since you say that
the server in your case is waiting 0.5 seconds, it's hard to say if
this is really "immediately" or if it is just a shorter delay than you

    I understand that servers will close idle connections after some
    time, but in this case a FIN packet arrives typically within 0.5
    seconds of the last successful request, which may be before or
    after the client has sent the next request.  The client is not
    pipelining requests - it waits for a response to each request
    before submitting the next - but does not expect the behaviour
    described above.  The servers in question are not stressed.

The client is *required* to expect that behavior -- see the MUST-level
requirement quoted above.

    If this behaviour is normal, can you confirm the natural corollary
    to this - that any HTTP/1.1 client must cope with a 'null' response
    to any 2nd or subsequent HTTP/1.1 request by creating a new socket
    and re-submitting the request - even where no 'connection: close'
    response header was included with the previous response?

Section 8.1.4 goes on to say:

				Client software SHOULD reopen the
   transport connection and retransmit the aborted sequence of requests
   without user interaction so long as the request sequence is
   idempotent (see section 9.1.2). Non-idempotent methods or sequences
   MUST NOT be automatically retried, although user agents MAY offer a
   human operator the choice of retrying the request(s). Confirmation by
   user-agent software with semantic understanding of the application
   MAY substitute for user confirmation. The automatic retry SHOULD NOT
   be repeated if the second sequence of requests fails.


P.S.: My personal belief is that a well-implemented server with
sufficient resources will try to keep a persistent connection
open for as long as possible.  The specification, however, does
not require this, and in many server architectures it seems to
be difficult to maintain large numbers of idle connections without
excessive resource consumption.
Received on Thursday, 11 October 2001 00:22:18 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:16:38 UTC