> I'm trying to think what sort of thing would return a RST under these
> conditions. MTU or duplex mismatches, or even SACK problems tend in my
> experience to give connections that starve themselves to death rather
> than being given a RST in midflow.
>
> Is the trace as you implied and the transfer running with decent
> throughput when a RST is received midtream?
> [As opposed to it stopping and then getting a RST after a while]
It's fast - here is an extract of the bit where it fails:
(leo is our mailhub - xxxxxx is the peer)
---------------------------------------------------------------------------
Packet 88
TIME: 14:44:37.461712 (0.061321)
TCP: xxxxxx.smtp -> leo.46621 Ack
DATA: <No data>
---------------------------------------------------------------------------
Packet 89
TIME: 14:44:37.461722 (0.000010)
TCP: leo.46621 -> xxxxxx.smtp Ack
DATA: m5hbCB2YWxpZGF0aW5nIGJvZGllcywg^M
Y291cnNlIHRpdGxlcyBmb3IgbmV3IGNvdXJzZXMgYXJlIGV2YWx1YXRlZCBm
^M
cm9tIGEgbWFya2V0IHBlcnNwZWN0aXZlB0lzIHRoZSBpbXBvcnRhbmNlIG9m
^M
<snip>
---------------------------------------------------------------------------
Packet 90
TIME: 14:44:37.461725 (0.000003)
TCP: leo.46621 -> xxxxxx.smtp Ack
DATA: gdGFrZW4g^M
b2YgZmFjdG9ycyBzdWNoIGFzIGNvc3RzLCBjb21wZXRpdGl2ZSBwcmljaW5n
^M
LCB0aGUgbWFya2V0knMgd2lsbGluZ25lc3Mgb3IgYWJpbGl0eSB0byBwYXks
^M
<snip>
---------------------------------------------------------------------------
Packet 91
TIME: 14:44:37.461728 (0.000003)
TCP: leo.46621 -> xxxxxx.smtp Ack
DATA: lIHByb21vdGlvbiBhbmQgb3V0cmVhY2guDQ1PdXRsaW5lIHRo^M
^M
bGwHQm9va2xldCCTUHJvY2VkdXJlcyBmb3IgdGhlIHByb3Bvc2FsIG9mIG5l
^M
dyBwcm9ncmFtbWVzIG9mIHN0dWR5lCBhZHZpc2VzIHN0YWZmIG9uIHRoZSB0
<snip>
---------------------------------------------------------------------------
Packet 92
TIME: 14:44:37.462376 (0.000648)
TCP: xxxxxx.smtp -> leo.46621 Rst
DATA: <No data>
---------------------------------------------------------------------------
Packet 93
TIME: 14:44:37.493309 (0.030933)
TCP: xxxxxx.smtp -> leo.46621 Ack
DATA: <No data>
---------------------------------------------------------------------------
Packet 94
TIME: 14:44:37.493324 (0.000015)
TCP: leo.46621 -> xxxxxx.smtp Rst
DATA: <No data>
---------------------------------------------------------------------------
Packet 95
TIME: 14:44:37.553997 (0.060673)
TCP: xxxxxx.smtp -> leo.46621 Ack
DATA: <No data>
---------------------------------------------------------------------------
One thing I'm not sure about is the Ack following the Rst from the remote
end - is this normal ?
Also, if I change the MTU using ifconfig - this change should be
effective immediately should it not ?
Regards and thanks,
Martin
