Problem
In my log files, I see an error such as:
Sep 13 14:52:58 machine pppd[pid]: Link id(339):dump bad accm packet
where "machine" and "pid" vary with the machine name and process id for the pppd process.
Is this error something to be concerned with, or can I safely ignore it?
Cause
The error indicates that a bad "Asynchronous Control Character Map" packet (ACCM packet) has been received through the PPP connection. That is, a packet containing bad or unintelligible serial line control information has been received.
PPP is run over serial lines using the RS232 protocol. This
protocol allows the use of control characters (^S, ^Q, and so on)
for flow control instructions to be sent back and forth between
the machines. If you are sending data that also has control
characters in it, those characters need to be encoded so they
aren't interpreted as flow control sequences.The ACCM field tells PPP which characters are being used by your hardware to control the connection so that it can manage those characters when they are part of packets.
Solution
Usually, if there are fewer than five or so of these errors, they
are safe to ignore. Most commonly, you will see the error when a
PPP connection is first brought up and the client and server are
negotiating link control options. In this early stage of PPP
negotiation, the client and server agree on maximum receive sizes,
authentication protocols, magic numbers, field compression, and
ACCM value.If you are using hardware flow control (rtscts), you can set
accm=0 in /etc/ppphosts and the control character sequences are
not mapped out. If either the client or the server have specific
settings required for a PPP connection, then both sides should be
manually set to agree on an ACCM value.In this case, assuming the PPP link is functioning satisfactorily, you can safely ignore bad accm errors in your log files.
If, however, the link is failing, you should turn debugging on
and watch the log files closely. Problems in the final IPCP stage
of PPP negotiation can be caused by mismatched parameters from
the earlier LCP stage, so make sure you resolve all "Nak"'s -
negative acknowledgements - found in the log file. See also
Other articles in this database, SCO OpenServer Networking Guide.