Is 2.4 wtmp/utmp still broken?

Is 2.4 wtmp/utmp still broken?

Post by Garrett Butul » Tue, 11 Apr 1995 04:00:00




(hostname = sparc) often indicates that a user has been logged
in and idle for hours, days, or even weeks.  However, `w` and
`finger` seem to work properly in the general sense (without
looking at a specific users), and only show users that are
currently logged in.

I thought I was current on finger/fingerd/wtmp/utmp patches,
but perhaps I'm not?  Or is this still par for the course?

Thanks.
-Garrett Butulis
 Wright State University
 Dayton, OH   USA

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by David Simmo » Wed, 12 Apr 1995 04:00:00





>(hostname = sparc) often indicates that a user has been logged
>in and idle for hours, days, or even weeks.  However, `w` and
>`finger` seem to work properly in the general sense (without
>looking at a specific users), and only show users that are
>currently logged in.

>I thought I was current on finger/fingerd/wtmp/utmp patches,
>but perhaps I'm not?  Or is this still par for the course?

I'm experiencing the same problem.  Does everybody have this problem,
or is it something that we are doing wrong?  I posted yesterday about
it, but only got a "me too" in the mail.  If anybody knows of a solution,
please post!!  Thanks.

David

--

Mississippi State University Electrical and Computer Engineering
Visit my home page!  http://www.ee.msstate.edu/~simmons

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by Stephen Anspa » Wed, 12 Apr 1995 04:00:00





: >(hostname = sparc) often indicates that a user has been logged
: >in and idle for hours, days, or even weeks.  

: I'm experiencing the same problem.  Does everybody have this problem,
: or is it something that we are doing wrong?  I posted yesterday about
: it, but only got a "me too" in the mail.  

  That'd be me (as in "me too").  Note that (at least on my site) console and OW
(e.g. pts/1, pts/2, etc.) logins are recorded correctly.  Logins via term/b (my
dialin modem) or telnet (on pts/0) are *not* marked as logged out until the next
person logs in on that terminal.

--
                                    - Steve -


Quis custodiet ipsos custodes?

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by Pegasus » Fri, 14 Apr 1995 04:00:00


The wtmp/utmp is still indeed broken in 2.4.  According to some Sun
literature (the source escapes me) the 2.4 versions of these services are
just the 2.3 versions with all of  the current patches integrated, much
like most of the 2.4 release of Solaris.

The 2.3 wtmp/utmp with all of the current patches applied to it behaves
just about the same way as it does in 2.4 (ie, it's not fully broken:
local finger and w work, but other functions just don't).

Many Sun representatives at both Sun and our master reseller, *and* the
Sun literature say that if you run 2.3 and you don't need Motif *and*
you've been keeping up to date on the patches from the SunSolve mailings,
you don't want to upgrade to 2.4.

Translation: wait until 2.5, or upgrade to 2.4 if you haven't been keeping
up with your patch sets (or if want to run Motif).

Kris

Pegasus Real-Time, Inc.
Opinions expressed by the Poster are the Poster's alone.

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by Derrick J. Brashea » Sat, 15 Apr 1995 04:00:00


Excerpts from netnews.comp.unix.solaris: 13-Apr-95 Re: Is 2.4 wtmp/utmp

Quote:> The wtmp/utmp is still indeed broken in 2.4.  According to some Sun
> literature (the source escapes me) the 2.4 versions of these services are
> just the 2.3 versions with all of  the current patches integrated, much
> like most of the 2.4 release of Solaris.

In what ways is it broken? I'm not disputing that it is, merely looking
for a fix for a particular problem we're having. That is, our local
version of login correctly writes a utmp entry. If you're telnetting,
telnetd un-writes it. If you're on console, nothing unwrites it. I'm
wondering what i need to do to convince something to unwrite it for
console logins

Thanks
-D

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by David Simmo » Sun, 16 Apr 1995 04:00:00




>That'd be me (as in "me too").  Note that (at least on my site) console and OW
>(e.g. pts/1, pts/2, etc.) logins are recorded correctly.  Logins via term/b (my
>dialin modem) or telnet (on pts/0) are *not* marked as logged out until the next
>person logs in on that terminal.

Okay... somebody suggested applying patch 102286-01.  This does indeed
fix the wtmp[x] problem.  'last' no longer shows a logged out session
as "still logged in".  However, the utmp[x] side of the bug persists.
A "finger -ms username" shows the terminated session as still being
logged in.  Does anybody know of a corresponding patch for this?

By the way, is there an authoritative ftp site for patches?  I couldn't
find 102286-01 from any of the usual places (sunsite, sunsolve1) and
had to go obtain the patch from my computing center.

David

--

Mississippi State University Electrical and Computer Engineering
Visit my home page!  http://www.ee.msstate.edu/~simmons

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by George Ro » Fri, 21 Apr 1995 04:00:00



> Many Sun representatives at both Sun and our master reseller, *and* the
> Sun literature say that if you run 2.3 and you don't need Motif *and*
> you've been keeping up to date on the patches from the SunSolve mailings,
> you don't want to upgrade to 2.4.

What about the scheduler changes...?
--
Dr George D M Ross, Department of Computer Science, University of Edinburgh
       Kings Buildings, Mayfield Road, Edinburgh, Scotland, EH9 3JZ

 
 
 

Is 2.4 wtmp/utmp still broken?

Post by Pegasus » Fri, 21 Apr 1995 04:00:00


We're not going to list all the reasons to upgrade to 2.4 and all the
reasons not to upgrade to 2.4 here.

Like the periodic sunsolve CDROMs, choose the features and/or bugfixes you
want on your system and evaluate whether you want to upgrade to 2.4 or
wait until 2.5.

Kris

Pegasus Real-Time, Inc.
Opinions expressed by the Poster are the Poster's alone.

 
 
 

1. xtacacs utmp/wtmp logging on Solaris 2.4

Hi!

I am using the Cisco 2511 terminalserver with xtacacs for the login
procedures etc. I have configured xtacacs to log utmp/wtmp and it seems
to do that fine, except for one slight problem; I can't seem to read the
files with the command last. After some "investigation" in the man files
I guess Solaris is using a different utmp/wtmp record, called utmpx/wtmpx
(Extended utmp/wtmp), and the command last seems to be expecting this type
of file.

So my questions goes out to all of you Solaris/xtacacs users, how do
you read them files later; i.e. for accounting etc?

Thankful for all replies.

Henrik Johansson
Internet One

--
  -----=<->=-----=<|>=-----=<+>=-----=<|>=-----=<->=-----=<+>=-----=<|>=-----
 /   Henrik Johansson    mobil: 0708-81 15 45   tel (arb): 031-775 00 80    /

-----=<->=-----=<|>=-----=<+>=-----=<|>=-----=<->=-----=<+>=-----=<|>=-----

2. FS: 42T 256MB/2GB/Key/Mouse/P200 Monitor

3. SUMMARY: Last, utmp, wtmp: a HIDEOUSLY BROKEN MESS

4. Virtual servers

5. Last, utmp, wtmp: a HIDEOUSLY BROKEN MESS

6. Ethernet on Dell Inspiron 5100

7. location of utmp wtmp in utmp.h

8. HLP: routing question (perhaps naive)

9. utmp and wtmp corruption...

10. How to remove utmp/wtmp entries?

11. utmp and wtmp re-init

12. Redhat 5.0 wtmp and utmp problems...

13. is it safe to simply truncate wtmp/utmp ??