Comment 74 for bug 1743249

On Tue, Feb 06, 2018 at 12:11:21AM -0000, Mike Pontillo wrote:
> Steve, can you be more specific about which packet capture showed the
> "stacked OACK" behavior?

This was the first packet capture that Jason posted, in comment #30. The
udp retransmits shown in packets 6262-6268 each receive an answering packet
in 6270-6271,6273-6277, in addition to 6269 as an answer to 6261. For
whatever reason, wireshark here does not decipher these duplicate OACK
packets as OACK, but an examination of the raw packets shows that's clearly
what they are.

> I looked at a packet capture Andres pointed me to, and don't see the
> "stacked OACKs" you describe. Each TFTP transaction (per RFC 1350) is
> indicated by the (source port, dest port) tuple, and I see that MAAS
> correctly OACKs each individual transaction (per RFC 2347) - not the
> retry packets within the same transaction.

Packets 6269-6271,6273-6277 are all answers to the same port on the client.
They don't have the same source port, because MAAS has allocated a separate
source port for each of these. It's not acking a separate individual
transaction, it's MAAS /creating/ a separate transaction (with the
allocation of a separate source port) for each one.

RFC2347 does not speak to this; the discussion of the port negotiation is in
RFC1350 §4:

   In order to create a connection, each end of the connection chooses a
   TID for itself, to be used for the duration of that connection. The
   TID's chosen for a connection should be randomly chosen, so that the
   probability that the same number is chosen twice in immediate
   succession is very low. Every packet has associated with it the two
   TID's of the ends of the connection, the source TID and the
   destination TID. These TID's are handed to the supporting UDP (or
   other datagram protocol) as the source and destination ports. A
   requesting host chooses its source TID as described above, and sends
   its initial request to the known TID 69 decimal (105 octal) on the
   serving host. The response to the request, under normal operation,
   uses a TID chosen by the server as its source TID and the TID chosen
   for the previous message by the requestor as its destination TID.
   The two chosen TID's are then used for the remainder of the transfer.

MAAS responds to 8 udp retransmits on srcport=25305, dstport=69 by sending 8
independent OACK packets back to dstport=25305 each from a different source
port.

Since Andres confirms that these duplicate acks still only result in one
database query, this may be a negligible bug if the only impact is duplicate
small udp packets. OTOH, depending on how MAAS implements this, it could
also result in port exhaustion on the server if unanswered OACKs are allowed
to linger.