12 Nov
2021
12 Nov
'21
10:43 p.m.
On Nov 12, 2021, at 4:18 PM, John Forecast <john at
forecast.name> wrote:
Yes, it appears in response to a NICE request. What's strange is that the connection
is accepted. If there's something wrong with the object, it shouldn't result in
a connect confirm but rather a connect reject (a.k.a., disconnect initiate with a suitable
reason code). The connect confirm tells the requesting end that the object is operational
and ready to accept requests. If something then goes wrong, that's ok but in that
case the requesting end expects an error reply according to the protocol. In this case
that would be some sort of NICE error message reporting a problem executing the request.
paul
On Nov 12, 2021, at 2:46 PM, Paul Koning
<paulkoning at comcast.net> wrote:
My network scanner is getting very strange responses from three different nodes all
misbehaving the same way: CAIR, MAGPIE, and FARGO.
What I see is:
1. The connection to object 19 (NICE, the network management listener) is accepted, but
the version number that is supposed to be sent as part of the accept is missing.
2. When I send a NICE request to the node, I get back a message with this content:
<28>Nov 12 20:35:16 dnetd[1688]: Cannot chdir to /nonexistent : No such file or
directory\n\xff\x00\x00
All three nodes do this. Configuration error that just happens to exist on all three?
paul
Those systems must be running DECnet for Linux. dnetd performs the mapping from an
inbound connect request and runs the appropriate daemon/user application. Assuming the
object database (/etc/dnetd.conf) has not been changed from the default, user ?nobody? has
a home directory of /nonexistent (which does not exist). The warning should be benign
since it will default running with the current directory of ?/?, although I?m not sure
that?s good - ?/tmp? may be a better default. I?m not sure where the extra hex values are
coming from. Are you seeing this as the reply to the NICE request? It should only be
written to syslog.
John.