Skip site navigation (1)Skip section navigation (2)
Date:      14 May 2002 08:57:04 +1000
From:      Andrew Reilly <areilly@bigpond.net.au>
To:        "Karl M. Joch" <k.joch@kmjeuro.com>
Cc:        stable@freebsd.org
Subject:   Re: 4.5 and mpd "no buffer space available on high load" ??
Message-ID:  <1021330624.8946.23.camel@gurney.reilly.home>
In-Reply-To: <3CDFEA8A.4070809@kmjeuro.com>
References:  <3CDFEA8A.4070809@kmjeuro.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 2002-05-14 at 02:32, Karl M. Joch wrote:
> Hi,
> 
> i have a fine working configuration at one customer:
> 
> 
>                               INET
>    |--------------------------------------------------------|
> 
>    |                                                        |
>    |                                                        |
>    |                                                        |
> -------                                               ----------
> BOX one			 
> 		        BOX two
> ADSL connection with mpd                          Leased Line connected
> IPSec Connection |===============================|IPSEC Connection
> Squid                                             Squid
> ipfw                                              ipfw
> Samba                                             Samba
>                                                    Wins Server (Samba)
> Cups                                              Cups
> ppp dialin server (for staff)                     ppp dialin server
> 
> 
> everything works great including wins browsing, printing via cups on 
> every location.........
> 
> there is a job running on box1 and box 2 every nigth sending out a ~20 
> MB file to the other location via nfs over the ipsec connection. the 
> jobs are planed in an way that first a sends to b and when this is 
> finished b sends to a. connection of box 1 is a 512/64 adsl in austria 
> and box 2 has a 512kb in germany.
> 
> box 2 can send the file just fine, but:
> 
> ->> when box 1 sends the file via cp command to box 2 (1 mounts 2 via 
> nfs) after it is created i immedialy get "no bufferspace available" on 
> every action box 1 does. the strange thing is, when letting the same 
> file being picked up by box 2 (box2 mounting box1 via nfs and running 
> the same cp command) it works just great.

This sounds similar to the problem that I've been having with a MS-PPTP
VPN link from my cable-connected FreeBSD box to the office Win2000 servr
(also connected by cable).  I posted to freebsd-questions a couple of
days ago.  [I wasn't doing nfs, but outgoing smtp and cvs messages would
sometimes just hang.]

I ran a script using jot to send ping packets across the link, with
sizes varying from 1300 to 2300 bytes, while also watching the link with
tcpdump.

Only one ping failed (it didn't even get out), with the following error
message:
ping: sendto: Message too long

I also saw a few "ip reassembly time exceeded" messages in tcpdump, but
I'm not sure whether they were correlated.

Most of my problems seem to have gone away when I added "add allow all
from any to any frag" to my firewall script.  Not sure how it ever
worked at all without that, though.

-- 
Andrew


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1021330624.8946.23.camel>