After re-building a server,
I can no longer rlogin as root.
I get the message in the subject line.
Seems the cure used to be in the inittab file but
running Solaris 2.6 now.
Any tips greatly appreciated.
I get the message in the subject line.
Seems the cure used to be in the inittab file but
running Solaris 2.6 now.
Any tips greatly appreciated.
> I get the message in the subject line.
> Seems the cure used to be in the inittab file but
> running Solaris 2.6 now.
> Any tips greatly appreciated.
> I get the message in the subject line.
> Seems the cure used to be in the inittab file but
> running Solaris 2.6 now.
> Any tips greatly appreciated.
# If CONSOLE is set, root can only login on that device.
# Comment this line out to allow remote login by root.
#
CONSOLE=/dev/console
See also man login.
Charles
--
Charles Soto - Austin, TX *** REMOVE "_ihatespam" from email address!
[insert annoyingly large ASCII art .sig here, and please, god, oh
please put an end to HTML in email and multi-part MIME!]
> After re-building a server,
> I can no longer rlogin as root.
> I get the message in the subject line.
> Seems the cure used to be in the inittab file but
> running Solaris 2.6 now.
> Any tips greatly appreciated.
>> After re-building a server,
>> I can no longer rlogin as root.
>> I get the message in the subject line.
>> Seems the cure used to be in the inittab file but
>> running Solaris 2.6 now.
>> Any tips greatly appreciated.
>"man login" should put you on the right track. You might also check
>/etc/default/login. Details escape me, but I see no reason to do your
>research for you, when the resources you need areon your machine
>waiting. The man pages are your friend.
CONSOLE=/dev/console
and comment it out
#CONSOLE=/dev/console
Jay
> >> After re-building a server,
> >> I can no longer rlogin as root.
> >> I get the message in the subject line.
> >> Seems the cure used to be in the inittab file but
> >> running Solaris 2.6 now.
> >> Any tips greatly appreciated.
> >"man login" should put you on the right track. You might also check
> >/etc/default/login. Details escape me, but I see no reason to do your
> >research for you, when the resources you need areon your machine
> >waiting. The man pages are your friend.
> In /etc/default/login, find the line
> CONSOLE=/dev/console
> and comment it out
> #CONSOLE=/dev/console
> Jay
maybe i don't understand the nature of the question since
my service provider rolls over news groups much too quickly
and i didn't see the original post.
essentially, by allowing root logins directly without being
on the system console or being some other validated user, you
really leave your system open to attacks. if you're going to
kick the 'security door' wide open then this is one of the
best ways to do it. especially (primarily?) if you're going
to hit the system as root across the bigger Internet.
if you insist on allowing root logins from anywhere then i
would strongly suggest the use of the ssh. at least that way
your root password doesn't transit the net 'in the clear'.
at any rate, i don't find logging in on my user account and
the su'ing to root that inconvenient. in fact, i *never* log
in as root (other than in system maintainance mode) directly.
think about what you're doing and for what reason - is it just
plain laziness? if so, you'll probably get what you ask for.
just a word of caution.
be careful,
dp
If root is also in /etc/ftpusers and the system is otherwise properlyQuote:>if someone has the root password, then /etc/default/login is NOT gonna
>stop him (her).
Note: Newsgroups seriously trimmed - I believe this is Solaris (and/or
SunOS) specific, as at least several other versions of Unix don't have
the same type of /etc/default/login mechanism - although many of them
have other, often similar ways, to implement effectively the same thing.
--
>> I agree that vi is not a good editor for passwd files. Primarily because
their can
>> be a
>> tendancy to walk away from your terminal, and finish up later.
>Which would apply to any editor, surely? I can't think off-hand of an
>editor which actually straps you to your chair until you've written the
>file away :-)
Adam
PS: No I do not want an editor war, the correct way to change password files
is to make the PFY do it.
>Avoid the situation happening in the first place, perhaps? Taking your
>"one in a million" comment seriously, assume a machine with 5000 active
>users (an ISP supplying shell accounts, perhaps). If these change their
>passwords once a week on average, this means 5000 password changes a
>week or 714 per day (assuming a 7 day week; make that 1000 per day if we
>are talking a 5 day working week). Even if these are spread evenly over
>a 24 hour day (which is unlikely) this is still an average of 29 per
>hour, or one write to the password file just over every two minutes.
>The chances of two changes colliding look somewhat less than "one in a
>million" given those odds, IMO.
Or copy the original passwd file off, make the changes to the bottom of theQuote:>Whatever the odds, why run the risk at all, when use of proper utilities
>to change password files will reduce the risk of colliding writes to
>approaching zero in any case?
That's true -- the passwd & shadow files are extremely important inQuote:>> If someone is so concerned over syntax checking and typos in their
passwd file,
>> you may want to
>> purchase Microsoft products instead of UNIX.
>I'm not entirely sure what you mean by this. Assuming the Unix machine
>is actually doing real work, and supporting a community of real users, I
>would submit that *any* system operator should be "so concerned" about
>the state of their password file!
That's why you save a copy of your passwd file. And back it up. And applyQuote:>Personally I don't care if my password file is completely hosed - I'm
>the only user on the machine on my desktop, I can easily boot from a
>network to repair it if I do hose it, most user information lives in
>NIS+ tables round here anyway, and at the worst I can always find a
>spare machine to use since my home directory lives on the network.
>The situation might well be different if the machine in question was a
>server supplying CPU to several hundred users, with no network booting
>available, no chance of downtime due to the thousands of dollars a
>minute lost by the business if the machine is down and no way to get
>into the machine as root because someone hamfisted hosed the password
>file.
In a well-run production environment, you'd undoubtedly have even more
safeguards.
You probably have a global passwd file anyway -- periodically pushed out via
rdist or
NIS or Tivoli or ... You keep that passwd file on a machine where it can be
tested
before it's pushed out.
I think the biggest reason to hand-edit the passwd file is that it works on
Solaris, Linux, HP-UX, etc. Otherwise, you have to remember the syntax
& flags on how to add users on each platform in your business.
Spike
Bottom line:
if someone has the root password, then /etc/default/login is NOT gonna
stop him (her).
regards,
Dave
+-----------------------------+----------------+Quote:>uuhhh.... but why?
>maybe i don't understand the nature of the question since
>my service provider rolls over news groups much too quickly
>and i didn't see the original post.
>essentially, by allowing root logins directly without being
>on the system console or being some other validated user, you
>really leave your system open to attacks. if you're going to
>kick the 'security door' wide open then this is one of the
>best ways to do it. especially (primarily?) if you're going
>to hit the system as root across the bigger Internet.
>if you insist on allowing root logins from anywhere then i
>would strongly suggest the use of the ssh. at least that way
>your root password doesn't transit the net 'in the clear'.
>at any rate, i don't find logging in on my user account and
>the su'ing to root that inconvenient. in fact, i *never* log
>in as root (other than in system maintainance mode) directly.
>think about what you're doing and for what reason - is it just
>plain laziness? if so, you'll probably get what you ask for.
>just a word of caution.
>be careful,
>dp
>Bottom line:
>if someone has the root password, then /etc/default/login is NOT gonna
>stop him (her).
>regards,
>Dave
On Mon, 17 May 1999 08:09:20 -0400,
>Because no matter how much you tell them not to do it some people still want
>to shoot themselves in the foot. The original poster asked a question, the
>second poster answered it. Is actually making the change stupid, yes but
>not matter how many times the reason for the security restriction is
>explained some people still insist on doing it. Same as some users still
>insist on editing /etc/passwd by hand and changing the root shell. People
>just get tired of telling them why not to do it.
Kai.
uuhhh.... why not editing /etc/passwd by hand?
This is what I always do. I just cannot remember if it's "useradd", or
"adduser" or "addUserAccount" or whatever on a certain flavour of Unix when
I want to add a
user. Isn't editing /etc/passwd by hand more consistent? (Perhaps I am
making a fool out of myself now, but I've never read I shouldn't do this
anywhere.)
Kai.
On some variants of 5.4.2 you can do what you like in the passwd file
andnothing happens until the next reboot or until you run the creatiadb
program, as passwd and shadow
info is held in the /etc/security/tcb master and index files, if you are
going to change things by hand then ,mae sure you do not leave yourself with
a duff passwd file!
Veltror
**** Posted from RemarQ - http://www.remarq.com - Discussions Start Here (tm) ****
> uuhhh.... why not editing /etc/passwd by hand?
> This is what I always do. I just cannot remember if it's "useradd", or
> "adduser" or "addUserAccount" or whatever on a certain flavour of Unix
> when I want to add a user. Isn't editing /etc/passwd by hand more consistent?
Potential problems (replacing "vi" with just about any hand editor, by
the way):
1) vi doesn't lock the password file. Consider the following scenario:
root does "vi /etc/passwd"
a user somewhere on the system changes their password using
"passwd"
root completes their changes and hits :wq
the user's changes will have been lost, since root has just overwritten
the
changed version of /etc/passwd with the version read in by vi before the
changes.
Even more interesting results will be obtained if the rewrite from
root's "vi" actually coincides with the rewrite from the user's "passwd"
invocation. At the worst the passwd file will be completely scrambled.
2) vi doesn't do any syntax or consistency checking. If you happen to
make a typo (one too many colons in a passwd file entry, for example),
nothing will pick it up until it's found that people can't log in.
I'd say you should at least use /usr/ucb/vipw instead of vi as vipw does
some syntax checking (it won't allow a garbled root passwd entry, for
example) and tries to lock the password file on entry to the editor.
Following a hand edit of the password file it's also a very good idea to
run pwck to check whether the file has been garbled. Do this BEFORE
exiting the root shell (as you might have problems su-ing to fix any
problems if the password file has, in fact, been damaged.
Preferably, however, you should use a utility that locks the password
file (and takes account of any locks that are already active on it),
syntax checks the password and shadow files, keeps passwd and shadow in
sync automatically, and checks consistency with other things such as
groups entries, existence of home directories, etc. The password file
is too important to take risks with, IMHO.
Regards
--
Tony
This posting is my own opinion and does not necessarily coincide with
the policy of Sun Microsystems. It does not constitute official support
from Sun Microsystems
1. GETSERVBYNAME()????????????????????"""""""""""""
Hi,
Does anyone know why
struct servent *serv;
serv=getservbyname("exec","tcp");
gives a warning err of incomparible pointer type?
I also can't get rexec to function. It compiles ok....
Thanks
Kirk
2. Horizontal screen lines with RH 5.0, ATI Mach64 CT, & Metro-X
3. "System too big" & details on console.c "Illegal asm or bug".
4. command logging/monitor shell
5. """"""""My SoundBlast 16 pnp isn't up yet""""""""""""
6. All programms do segfault after a while
7. "Standard Journaled File System" vs "Large File Enabled Journaled File System"
9. "Interrupted system call" at "low level" - system calls
10. Dynix "Not on system console" with root login
12. RedHat 5.2, "/boot/System.map" and "/boot/System.map"
13. Type "(", ")" and "{", "}" in X...