[5756] in Kerberos
Re: telnetd and utmp in K5.5
daemon@ATHENA.MIT.EDU (Sam Hartman)
Tue Aug 22 23:28:05 1995
To: ramus@nersc.gov (Joe Ramus)
Cc: kerberos@MIT.EDU
In-Reply-To: Your message of "Tue, 22 Aug 1995 17:47:40 PDT."
<9508230047.AA00839@windsail.nersc.gov>
Date: Tue, 22 Aug 1995 23:20:17 EDT
From: Sam Hartman <hartmans@MIT.EDU>
>>>>> "Joe" == Joe Ramus <ramus@nersc.gov> writes:
Joe> I am a bit puzzled about the behavior of telnetd regarding
Joe> the utmp file. This is for Kerberos 5.5.
Joe> On our Solaris 2.4 systems, it looks like there are two
Joe> entries in the utmp file for each login when we do the
Joe> Kerberos authenticated login. One entry is created by
Joe> telnetd and the second is created by login.krb5.
Joe> It seems to me that the correct behavior is to just have
Joe> login.krb5 do the utmp update.
Joe> ----------------------------------------------------------------
Joe> | Joe Ramus NERSC Livermore (510) 423-8917 ramus@nersc.gov |
Joe> ----------------------------------------------------------------
No, this is not correct. Telnetd should create a login-type
utmp entry, and if this is done properly, login.krb5 will re-use the
same slot so that the session is tracked as soon as connected, and the
user name is filled in on login. If rlogind/login.krb5
orlogind/system -supplied login fail to do this properly, it is a bug.
We don't care about the behavior of beta-5 telnetd at this point as
that code has been replaced.
--Sam
