[Spread-users] SP_connect() and -6 REJECT_NOT_UNIQUE with remote daemons

[Alexey Zakhlestin]

just a guess: modern tcp stacks can keep connection "open" even during
periods of problems on lower-level. good example is ssh-console, which
stays in working state after you plug-out and then plug-in cable.

So, could it be the case, that physical link has problems (and you
detected it), but logical connection is still ok, so reconnection is
not really needed?

On 6/13/08, Matt Garman <matthew.garman at gmail.com> wrote:
>
>  We experience the following scenario from time-to-time:
>
>     - When using a spread program (client) with a remote daemon, the
>       link between the client and daemon temporarily goes down.
>       (I.e. network problems)
>
>     - Our client programs are written to detect a disconnect, and
>       they automatically try to re-connect.
>
>     - But when the link is restorted, the connection fails with:
>
>         SP_error: (-6) Connection rejected, name not unique
>
>  Our assumption is that when the connection goes down, the daemon
>  doesn't properly register the disonnect.  So it keeps all of the
>  user names as active.  Then when the connection comes back, and the
>  processes try to reconnect, the daemon rejects because it thinks the
>  names are already used.
>
>  The API docs suggest that we try connecting again with a different
>  name.  This isn't really practical for us... Is there any other way
>  to force the daemon to recognize the client disconnect?
>
>  Thanks,
>  Matt
>
>
>  _______________________________________________
>  Spread-users mailing list
>  Spread-users at lists.spread.org
>  http://lists.spread.org/mailman/listinfo/spread-users
>


-- 
Alexey Zakhlestin
http://blog.milkfarmsoft.com/