for the next edition of host requirements

Bob Braden <braden@isi.edu> Fri, 21 January 1994 01:47 UTC

Received: from ietf.nri.reston.va.us by IETF.CNRI.Reston.VA.US id aa09605; 20 Jan 94 20:47 EST
Received: from CNRI.RESTON.VA.US by IETF.CNRI.Reston.VA.US id aa09601; 20 Jan 94 20:47 EST
Received: from venera.isi.edu by CNRI.Reston.VA.US id aa17522; 20 Jan 94 20:47 EST
Received: from zephyr.isi.edu by venera.isi.edu (5.65c/5.61+local-14) id <AA01344>; Thu, 20 Jan 1994 17:32:42 -0800
Received: by zephyr.isi.edu (5.65c/5.61+local-16) id <AA00802>; Thu, 20 Jan 1994 17:32:40 -0800
Date: Thu, 20 Jan 1994 17:32:40 -0800
Sender: ietf-archive-request@IETF.CNRI.Reston.VA.US
From: Bob Braden <braden@isi.edu>
Message-Id: <199401210132.AA00802@zephyr.isi.edu>
To: postel@isi.edu
Subject: for the next edition of host requirements
Cc: braden@isi.edu, ietf-hosts@isi.edu

Jon,

I agree with your response, with the additional comment that an FTP
should do whatever an application normally does to ensure the file
has reached stable storage; it is not required to take heroic
measures.  This refers to a Unix-like file system, which may
not actually write a file to disk ("sync") until later.  We
put up with the unreliability that implies in order to get
the performance, and it should be good enough for FTP.  But it
would NOT be nice to give a 2xx reply before writing out the last
block and closing the file!!!

Bob

----- Begin Included Message -----

From postel Thu Jan 20 16:47:22 1994
Date: Thu, 20 Jan 1994 16:47:22 -0800
From: postel (Jon Postel)
To: braden
Subject: for the next edition of host requirements
Cc: postel
Content-Length: 2323
X-Lines: 80


----- Begin Included Message -----

>From ptrei@bistromath.mitre.org Wed Jan 19 11:29:27 1994
Posted-From: The MITRE Corporation, Bedford, MA
To: jkrey@ISI.EDU, postel@ISI.EDU
Cc: ptrei@mitre.org
Subject: Quickie FTP protocol question.
Date: Wed, 19 Jan 94 14:28:26 -0500
From: "Peter G. Trei" <ptrei@bistromath.mitre.org>


Hello:
     I hope I've got the right people: the authors of RFC 959, the FTP
protocol spec. If not, please ignore this mail.

Scenario:
Host A sends a file to host B.
The transfer completes.
Host A recieves an acknowledgement of it's EOF marker.

Did host B write the end of the transfered file to disk before sending
the eof-ACK to host A? Is this required by the protocol, or is it left
unspecified? I can't see anything in the RFC which requires it.

A one-line answer would be plenty.
						thanks,
							Peter Trei
							ptrei@mitre.org
----- End Included Message -----


----- Begin Included Message -----

>From postel Thu Jan 20 16:45:39 1994
Date: Thu, 20 Jan 1994 16:45:34 -0800
From: postel (Jon Postel)
To: ptrei@bistromath.mitre.org
Subject: Re: Quickie FTP protocol question.
Cc: postel, jkrey


Peter:

I would say that the general rule that a 200 series reply means that
the requested action is completed would imply that if the command 
was STOR the file was safely stored of the destination system and
would not be lost if the system crashed at that point.

By the way, you should be reading RFC-1123 for further details on FTP.

--jon.

   To: jkrey@ISI.EDU, postel@ISI.EDU
   Cc: ptrei@mitre.org
   Subject: Quickie FTP protocol question.
   Date: Wed, 19 Jan 94 14:28:26 -0500
   From: "Peter G. Trei" <ptrei@bistromath.mitre.org>
   
   
   Hello:
        I hope I've got the right people: the authors of RFC 959, the FTP
   protocol spec. If not, please ignore this mail.
   
   Scenario:
   Host A sends a file to host B.
   The transfer completes.
   Host A recieves an acknowledgement of it's EOF marker.
   
   Did host B write the end of the transfered file to disk before sending
   the eof-ACK to host A? Is this required by the protocol, or is it left
   unspecified? I can't see anything in the RFC which requires it.
   
   A one-line answer would be plenty.
   						thanks,
   							Peter Trei
   							ptrei@mitre.org
   
----- End Included Message -----



----- End Included Message -----