BackupPC-users

Re: [BackupPC-users] All incr fail(SIGPIPE) all full complete - cygwin/W7

2015-09-11 15:50:56
Subject: Re: [BackupPC-users] All incr fail(SIGPIPE) all full complete - cygwin/W7
From: Les Mikesell <lesmikesell AT gmail DOT com>
To: "General list for user discussion, questions and support" <backuppc-users AT lists.sourceforge DOT net>
Date: Fri, 11 Sep 2015 14:49:17 -0500
On Fri, Sep 11, 2015 at 11:55 AM,  <fred AT damen DOT org> wrote:
>>>> 2015/09/10 18:35:35 [7572] rsync: [sender] write error: Connection reset by
>>> peer (104)
>>> 2015/09/10 18:35:35 [7572] rsync error: error in socket IO (code 10) at
>>> io.c(820) [sender=3.1.1]
>>>
>>> Form the Wireshark log on the backuppc server(linux) during this time frame
>>> it
>>> appears as though the client(172.29.35.65)'s rsync is sending a RST(reset),
>>> packet no. 4918.
>>
>> That scenario is pretty common when there is a nat gateway involved,
>> but a stateful host firewall could also time out the connection and
>> start blocking after some amount of idle time.
>
> The networking device connecting the two computers is a highly configurable
> Cisco router and I am told that it was configured to act as a simple switch
> between the computers on the LAN.  I would not be surprised if the traffic is
> routed up through the router and back down through the router between the
> server and client.  Do you have references or what can I google to get more
> info on this?  I will need ammo to convince the administrator of the router
> that it is their problem.  I will try connecting server and client using a
> 'simple' switch.

If the IP addresses of the source and target both fit within the
netmask, the router should not be involved in the connection.  The
addresses you mention aren't within a typical class C range (/24), so
that would depend on your setup.

> I do not suspect that the firewall on the server and/or client as there
> appears to be KeepAlive messages going back and forth.  Three minutes seems to
> be a short time for a timeout.  Why does the client rsync appear to be sending
> a reset?

There's really no reason for the client rsync to send a reset.
Normally that is an os/socket level action indicating that the
destination application target has closed the socket or nothing is
listening on the destination port.   It could mean the application
crashed and exited unexpectedly.   Or it can be from an intermediate
network device or a software host firewall configured to reject (not
drop) packets in certain circumstances - like 'established'
connections with an idle timeout.     Keepalives 'should' keep
timeouts from happening, though.

> How can this timeout be changed?  I did not see any option for this
> when I setup the firewall rules for the rsync connection.  Also I remember
> testing this once with the firewall on the client shutoff, with the same
> results.

I don't understand it well enough to give any advice.    Have you
watched the client side with wireshark to see if the client is
actually receiving anything when the rst is sent?

-- 
  Les Mikesell
     lesmikesell AT gmail DOT com

------------------------------------------------------------------------------
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/

<Prev in Thread] Current Thread [Next in Thread>