jr@oglvee.UUCP (Jim Rosenberg) (12/13/89)
We are suffering from a recurring arthritis of the loginism. The symptoms range from pain in the outer mail utilities all the way to collapse of the HDB validation. It seemed pretty clear something was going wrong with /etc/utmp, and sure enough that proved to be the case. We keep getting (or should I say gettying?) just a couple or three tty lines with more than one entry in /etc/utmp. Here's an example: LOGIN tty12 Dec 11 18:24 . 11884 jr tty12 Dec 12 08:19 . 11884 The above two lines were grepped from a who -a listing. I *thought* /etc/utmp was only supposed to have one entry per tty. It appears that utilities don't all do the same thing regarding utmp: Some assume that the first matching entry is correct, others take the trouble to search the whole file. Elm, for instance, has no trouble with this situation and realizes that I'm jr, not LOGIN. But the elm utilities wnewmail and from will no longer work right when the second utmp entry for tty12 materializes. (This is elm 2.1, 2.2 may fix this for all I know.) Our version of HDB messes up badly under these conditions. When I get a second utmp entry for the modem line, it appears in doing site validation HDB is pulling off the login name from the first (WRONG!) utmp entry. It concludes the user is LOGIN. But then it seems to know this can't be right (no passwd entry for LOGIN?) so it paws through /etc/passwd for the first entry that matches by uid. Not fun at all if you have multiple uuwhoever logins all with the same uid as uucp. So when the modem line acquires a callus of the utmpnail HDB validation suddenly starts failing. This is really annoying. I've fixed it by going single-user and catting /dev/null >/etc/utmp, but now it's come back. Does anybody know how this happens? Our system is an Altos 2000 with Altos UNIX 5.3.1 d. [Please excuse if this is one of those oft-discussed questions; I've been buried under a mound of MIS work lately & haven't had much time to read news.] -- Jim Rosenberg pitt Oglevee Computer Systems >--!amanue!oglvee!jr 151 Oglevee Lane cgh Connellsville, PA 15425 #include <disclaimer.h>
allbery@ncoast.org (Brandon S. Allbery) (12/14/89)
In article <519@oglvee.UUCP> jr@oglvee.UUCP (Jim Rosenberg) writes:
LOGIN tty12 Dec 11 18:24 . 11884
jr tty12 Dec 12 08:19 . 11884
That is most likely a bug in login. Are you running the stock 5.3d login, or
one gleaned from the Usenet? Or a non-stock getty?
I've not seen this on other systems running Altos 5.3dS1, including one 2000.
(If you're running dS0, consider getting the update for dS1.)
++Brandon
--
Brandon S. Allbery allbery@NCoast.ORG, BALLBERY (MCI Mail), ALLBERY (Delphi)
uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu bsa@telotech.uucp
*(comp.sources.misc mail to comp-sources-misc[-request]@backbone.site, please)*
*Third party vote-collection service: send mail to allbery@uunet.uu.net (ONLY)*
expnet.all: Experiments in *net management and organization. Mail me for info.
--
Brandon S. Allbery allbery@NCoast.ORG, BALLBERY (MCI Mail), ALLBERY (Delphi)
uunet!hal.cwru.edu!ncoast!allbery ncoast!allbery@hal.cwru.edu bsa@telotech.uucp
*(comp.sources.misc mail to comp-sources-misc[-request]@backbone.site, please)*
*Third party vote-collection service: send mail to allbery@uunet.uu.net (ONLY)*
expnet.all: Experiments in *net management and organization. Mail me for info.rob@PacBell.COM (Rob Bernardo) (12/14/89)
In article <519@oglvee.UUCP> jr@oglvee.UUCP (Jim Rosenberg) writes:
+ It seemed pretty clear something was going wrong with
+/etc/utmp, and sure enough that proved to be the case. We keep getting (or
+should I say gettying?) just a couple or three tty lines with more than one
+entry in /etc/utmp. Here's an example:
...
+But the elm utilities wnewmail and from will no longer work right when
+the second utmp entry for tty12 materializes. (This is elm 2.1, 2.2 may fix
+this for all I know.)
This made shed some light on the matter.
The elm utility wnewmail (version 2.2) is a program that runs for the duration
of a login session. It knows to die when one of the following happen:
1. it receives a hangup signal, as is sent to all processes of
of terminal's process group when the user logs out.
2. getppid() returns 1, which is to say that the originating login
shell has gone away and the wnewmail process has taken on init
as its parent.
3. isatty(1) returns 0, which is to say standard output is not
a terminal (I don't quite see why this would *change* over the
course of the process running and it may only be a test that
would fail the first time it's tested).
It's possible that version 2.1 of wnewmail dies under #1 above as #2 and #3
were added in some recent version of ELM maybe after 2.1.
In any case, I presume that by your saying "wnewmail no longer works
right when the second utmp entry materializes", you mean wnewmail
terminates prematurely. Since wnewmail does not consult the utmp file
at all, then you may have a situation of whatever is wrong causing both the
utmp phenomenon and the wnewmail phenomenon separately, *not* a situation of
the utmp phenomenon causing the wnewmail phenomenon.
--
Rob Bernardo ...![backbone]!pacbell!pbhyf!rob -or- rob@pbhyf.PacBell.COM
Product engineer, UNIX/C Reusable Code Library Editor, "Go `C' UNIX"
Office: (415) 823-2417 Pacific * Bell, San Ramon, California
Residence: (415) 827-4301 R BAR JB, Concord, California