[OpenAFS-devel] rx abort, again

[Jeffrey Hutzelman]

On Wednesday, October 05, 2005 11:52:30 AM -0400 Jim Rees <rees@umich.edu> 
wrote:

> I thought we had fixed the problem where a client ignores an rx abort
> from a fileserver, but now I'm seeing it on a client built from cvs head
> yesterday.
>
> Shouldn't the client tear down the connection and start a new one?  Or
> does abort just apply to that call, not the entire connection?

Disclaimer: I've not looked at your trace yet.

An Rx abort packet is the mechanism used to deliver all error returns from 
RPC's, regardless of the reason.  As such, they normally do _not_ apply to 
the entire connection, but rather only to a single call.

In certain cases, aborts may be sent which apply to an entire connection. 
For example, if a security class signals an error, that is generally a 
connection-level error.  The bug we fixed was one where new calls created 
by calling rx_NewCall on a connection already in an error state did not 
inherit the error.  The result was that Rx would attempt to send traffic on 
such a call, instead of simply returning the same error code to the caller.

Until I look at the trace, I won't be able to easily tell if you're seeing 
the same problem or something different.

-- Jeff