Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Groups keyboard shortcuts have been updated
Dismiss
See shortcuts

Timeout during message collect

210 views
Skip to first unread message

Lch

unread,
Sep 14, 2002, 10:53:58 AM9/14/02
to

I have a RedHat 7.2 box running among other things sendmail.
I sometimes get these weird errors in the daily logwatch when people try to
send large messages to my users.

--------------------- sendmail Begin ------------------------

3180920 bytes transferred
28 messages sent

**Unmatched Entries**

g8CKKt019090: timeout waiting for input from mail.olavolsen.no during
message collect
g8D8HB020759: ruleset=check_mail, arg1=username@put_a_valid_domain.com,
relay=[211.238.143.84], reject=553 5.1.8 username@put_a_valid_domain.com...
Domain of sender address username@put_a_valid_domain.com does not exist
g8DBIG021171: timeout waiting for input from mail.olavolsen.no during
message collect
g8DBSP021205: timeout waiting for input from mail.olavolsen.no during
message collect
g8DBUv021216: timeout waiting for input from mail.olavolsen.no during
message collect
g8DBmk021245: timeout waiting for input from mail.olavolsen.no during
message collect
g8DC16021276: timeout waiting for input from mail.olavolsen.no during
message collect
g8DCM0021316: timeout waiting for input from mail.olavolsen.no during
message collect
g8DGSE021760: timeout waiting for input from mail.olavolsen.no during
message collect

---------------------- sendmail End -------------------------

I takes up to 8-10 hours to get the messages delivered...
Does anyone have a clue on whats going on?
Is this a line speed issue? I have a 700/128 kBit ADSL line, with fixed IP
for the NAT-router.
Smaller messages works perfectly...

Lch


Andrzej Filip

unread,
Sep 14, 2002, 5:11:02 PM9/14/02
to
Lch wrote:

You may have encountered yet another effect of problems described in
http://www.sendmail.org/faq/section3.html#3.10
Q3.10 -- How do I solve "collect: I/O error on connection" or "reply: read
error from host.name" errors?

In short:
when the remote site sends you "max size" TCP packet (after DATA command) the
packets may be lost without a trace.

Could you post MSS of your default route ?
[ TCP packet payload size ]
AFAIK RFC recomends MSS 576 for default route but RedHat uses 1500 (tuned for
ethernet).

Use the commands below to get MSSes:
netstat -nr
OR
route -ne

--
Andrzej [pl>en: Andrew] Adam Filip http://www.polbox.com/a/anfi/
I may disagree with the following *random epigram* :
I came out of twelve years of college and I didn't even know how to sew.
All I could do was account -- I couldn't even account for myself.
-- Firesign Theatre

Lch

unread,
Sep 15, 2002, 5:08:32 AM9/15/02
to

"Andrzej Filip" <an...@Box43.pl> wrote in message
news:3D83A5E6...@Andrzej.Adam.Filip...

Okay, here goes, I ran the two commands on the RedHat box:

Kernel IP routing table
Destination Gateway Genmask Flags MSS Window irtt
Iface
192.168.0.0 0.0.0.0 255.255.255.0 U 40 0
0 eth0
127.0.0.0 0.0.0.0 255.0.0.0 U 40 0
0 lo
0.0.0.0 192.168.0.1 0.0.0.0 UG 40 0
0 eth0

It appears that MSS is set to 40 for all my connections. The optimum is 576
or 1500?
How do I change this? Do I have to do that for all the machines I have
communicating with the Linux machine on my local lan (192.168.0)?

I apprechiate your help, thanx! :)

Lch


Andrzej Filip

unread,
Sep 15, 2002, 10:23:11 AM9/15/02
to
Lch wrote:

> "Andrzej Filip" <an...@Box43.pl> wrote in message
> news:3D83A5E6...@Andrzej.Adam.Filip...
> > Lch wrote:
> >
> > > I have a RedHat 7.2 box running among other things sendmail.
> > > I sometimes get these weird errors in the daily logwatch when people try
> to
> > > send large messages to my users.

> > >[...]


> > > g8CKKt019090: timeout waiting for input from mail.olavolsen.no during
> > > message collect

> > >[...]


> > > I takes up to 8-10 hours to get the messages delivered...
> > > Does anyone have a clue on whats going on?
> > > Is this a line speed issue? I have a 700/128 kBit ADSL line, with fixed
> IP
> > > for the NAT-router.
> > > Smaller messages works perfectly...
> >
> > You may have encountered yet another effect of problems described in
> > http://www.sendmail.org/faq/section3.html#3.10
> > Q3.10 -- How do I solve "collect: I/O error on connection" or "reply: read
> > error from host.name" errors?
> >
> > In short:
> > when the remote site sends you "max size" TCP packet (after DATA command)
> the
> > packets may be lost without a trace.
> >
> > Could you post MSS of your default route ?
> > [ TCP packet payload size ]
> > AFAIK RFC recomends MSS 576 for default route but RedHat uses 1500 (tuned
> for
> > ethernet).
> >
> > Use the commands below to get MSSes:
> > netstat -nr
> > OR
> > route -ne
> >
>

> Okay, here goes, I ran the two commands on the RedHat box:
>
> Kernel IP routing table
> Destination Gateway Genmask Flags MSS Window irtt Iface
> 192.168.0.0 0.0.0.0 255.255.255.0 U 40 0 0 eth0
> 127.0.0.0 0.0.0.0 255.0.0.0 U 40 0 0 lo
> 0.0.0.0 192.168.0.1 0.0.0.0 UG 40 0 0 eth0
>
> It appears that MSS is set to 40 for all my connections. The optimum is 576
> or 1500?

In this case 40 means 0 [mss is set per device used - 1500 for ether net card].

Linux kernel keeps mss without TCP header size (40 bytes), route reports mss
with TCP header size included. It needlessly converts 0 to 40 :)
[ linux_kernel_mss + 40 = route_reported_mss ]

1500 is optimum value for ether net (LAN) transmission. AFAIR 576 is
(RFC?) recommended value for default ip route.

> How do I change this?

First try if fixing mss for mail.olavolsen.no connection only fixes your
problem

route add mail.olavolsen.no mss 576 gw 192.168.0.1

see "man route" for more details

> Do I have to do that for all the machines I have
> communicating with the Linux machine on my local lan (192.168.0)?

How mail.olavolsen.no is connected ? (LAN, dialup WAN, internet)

> I apprechiate your help, thanx! :)

How to make mss change for default route persistent (reboot resistant) is a
good question for red hat related news group.

[I did it a few years/releases ago on my system]

--
Andrzej [pl>en: Andrew] Adam Filip http://www.polbox.com/a/anfi/
I may disagree with the following *random epigram* :

It is better to live rich than to die rich.
-- Samuel Johnson

Lch

unread,
Sep 16, 2002, 4:52:42 PM9/16/02
to
It works perfectly now, at least from the server in question...
I havent seen the issue from any other server - yet... :)

Thanx a lot

Lch

"Andrzej Filip" <an...@Box43.pl> wrote in message

news:3D8497CF...@Andrzej.Adam.Filip...

Andrzej Filip

unread,
Sep 17, 2002, 2:45:53 PM9/17/02
to
Lch wrote:

> "Andrzej Filip" <an...@Box43.pl> wrote in message

> > [ how to fix MSS using "per host" route on linuxes]


> It works perfectly now, at least from the server in question...
> I havent seen the issue from any other server - yet... :)

Do not forget to make the route "reboot resistant" :)

--
Andrzej [pl>en: Andrew] Adam Filip http://www.polbox.com/a/anfi/
I may disagree with the following *random epigram* :

If there is no wind, row.
-- Polish proverb

0 new messages