webadmin-devel — The webmin development mailing list

Re: creating a new module

[Jamie C. <jca...@we...>]

Jannetta S Lewis wrote:

> Could someone perhaps assist a dumb newbie. I am working my way through
> the documentation on creating a new module for webmin. I created the
> directory, the images dir, the index.html and the modules.info file.
> However, despite the fact that I have a desc=description line in
> modules.info, the name of the module does not appear under the icon. What
> is wrong?


The file should be called module.info, not modules.info . Also, because webmin
caches this file you need to delete /etc/webmin/module.infos.cache every time
you modify it, so that the cache is re-loaded.

  - Jamie

creating a new module

[Jannetta S L. <jan...@he...>]

Could someone perhaps assist a dumb newbie. I am working my way through
the documentation on creating a new module for webmin. I created the
directory, the images dir, the index.html and the modules.info file.
However, despite the fact that I have a desc=description line in
modules.info, the name of the module does not appear under the icon. What
is wrong?

================================================================
Jannetta S Lewis
email: jan...@he...
home page: http://bright-ideas.keystroke.info
================================================================

Re: Bug on lpadmin module

[Jamie C. <jca...@we...>]

David Chow wrote:

> Hi all,
> 
> A bug found on creating a printer entry using the lpadmin module. Please 
> notice the -sDEVICE=[device] is incorrect and result gs to complain. 
> Please correct this problem. Thanks.
> 
> The file content is from /etc/webmin/lpadmin/drivers/test (My printer 
> name is called test with the driver "ljet4" )


Thanks, I have found the cause of this .. an update to fix it has been
released at http://www.webmin.com/updates.html

  - Jamie

Bug on lpadmin module

[David C. <dav...@sh...>]

Hi all,

A bug found on creating a printer entry using the lpadmin module. Please 
notice the -sDEVICE=[device] is incorrect and result gs to complain. 
Please correct this problem. Thanks.

The file content is from /etc/webmin/lpadmin/drivers/test (My printer 
name is called test with the driver "ljet4" )

#!/bin/sh
# Name: test
# Type: HP LaserJet 1100   (ljet4)
# DPI: 300
LD_LIBRARY_PATH=
PATH=/bin:/usr/bin:/sbin:/usr/sbin:/usr/local/bin
export LD_LIBRARY_PATH PATH GS_FONTPATH GS_LIB
/usr/bin/perl -e 'while(<STDIN>) { print if (!/^\s*#####/); }' >/tmp/$$.gsin
gs -sOutputFile=/tmp/$$.gs -dSAFER -sDEVICE= -r300 -dNOPAUSE 
/tmp/$$.gsin </dev/null >/dev/null 2>&1
rm /tmp/$$.gsin
cat /tmp/$$.gs
rm /tmp/$$.gs

The correct version of the file should be following and I have tested to 
be work.

#!/bin/sh
# Name: test
# Type: HP LaserJet 1100   (ljet4)
# DPI: 300
LD_LIBRARY_PATH=
PATH=/bin:/usr/bin:/sbin:/usr/sbin:/usr/local/bin
export LD_LIBRARY_PATH PATH GS_FONTPATH GS_LIB
/usr/bin/perl -e 'while(<STDIN>) { print if (!/^\s*#####/); }' >/tmp/$$.gsin
gs -sOutputFile=/tmp/$$.gs -dSAFER -sDEVICE=ljet4 -r300 -dNOPAUSE 
/tmp/$$.gsin </dev/null >/dev/null 2>&1
rm /tmp/$$.gsin
cat /tmp/$$.gs
rm /tmp/$$.gs

Just fill up the -sDEVICE=ljet4 or with the correct device driver name 
and it will work correctly. This bug is reproducible through create a 
printer entry under the lpadmin webmin module.

regards,
David

Re: Themes with shared icon set, symbolic links?

[Jamie C. <jca...@we...>]

Joe Cooper wrote:

> Hey Jamie,
> 
> I have two themes that share an icon set.  Using a symbolic link seems 
> to work, but I've seen enough path issues pop up on the list to know to 
> be cautious.
> 
> Any reason you can think of that linking all of the module directories 
> to another theme to grab the icons will break something subtle?


That will work fine, as long as the linked-to theme is always installed.
In fact, I have done it myself in some cases ..

  - Jamie

Themes with shared icon set, symbolic links?

[Joe C. <jo...@sw...>]

Hey Jamie,

I have two themes that share an icon set.  Using a symbolic link seems 
to work, but I've seen enough path issues pop up on the list to know to 
be cautious.

Any reason you can think of that linking all of the module directories 
to another theme to grab the icons will break something subtle?
--
Joe Cooper <jo...@sw...>
Web caching appliances and support.
http://www.swelltech.com

Re: Mgetty

[Jamie C. <jca...@we...>]

Jami wrote:

> I am getting a message saying Mgetty is not installed?  is that a 
> separate package, and do I need it?


If you want to set your system up as a dial-in server, yes you need it.
Most linux distributions include it as a package though.

  - Jamie

Mgetty

[Jami <pc...@it...>]

I am getting a message saying Mgetty is not installed?  is that a =
separate package, and do I need it?

[Usermin] Ability to logout gone?

[Tony S. <tsh...@st...>]

I'm in the process of migrating servers, and part of that was builing a new
system and installing the latest FreeBSD, Webmin, and Usermin.

I migrated our custom theme over, and logged in as myself.  All looked well,
but I wanted to test another user....however the printed link (I'm using a
modified version of gehrigal's usermin theme) that used to be on the left is
gone!  I went through the options page on webmin, and for some reason I
can't find anything under Authentication related to this.

I'm sure I'm overlooking something simple...

I hope?

Tony Shadwick
Manager of Internet Services
Strategic Technology Group
314-480-1324 

Re: &error() and locks...

[<jam...@te...>]

> Unfortunately, miniserv already defines an END function that is uses for
> a different purpose .. Is it possible to have more than one though, in
different
> packages?

Yeah, you can do that.  It is using is probably using at_exit() under the
hood to register
callback to be performed at exit().  These callbacks are put on stack of
functions
to be called at exit time.

The one issue I see is that you are using the "do" or "require" statement
to pull in
your libraries, so what happens is that though you would end up calling the
END routine, since the "require" is doing everything in the same namespace
as the
script when the calling script's END subroutine gets parsed it redefines
the one that
was created in the library.  So unfortunatley, doing what I said is not
possible
at first glance (it would be if web-lib.pl was web-lib.pm, and called as
use web-lib).

You can still get there with your current architecture though.  What you
would have to
do is create a function in the web-lib.pl, that registers callbacks to be
ran at
exit().  It would create a global (could be lexically private with file
scope)
list of callbacks.  This list would then get ran in web-lib.pl's END
function.
Then the END fucntion in the miniserv.pl program would just need a name
change, and
be registered via the registration function.  Here is a quick dirty
example:

      <<<web-lib.pl>>>
     #
     # Instantiate the private END callback stack
     my @END_callbacks = ();

     sub register_END
     {
     my $call_back = shift;

     push(@END_callbacks, $call_back);
     }

     sub END
     {
     #
     # Run callbacks
     foreach my $callback (@END_callbacks) {
          &{$callback}();
          }

     #
     # Do all the other web-lib things...
     }


     <<<miniserv.pl>>>
     #
     # Somewhere at the begining you register your miniserv.pl end
     # function.
     register_END(\&miniserv_END);

Cheers...james

Re: &error() and locks...

[Jamie C. <jca...@we...>]

jam...@te... wrote:

>>>Hi Jamie,
>>>
>>>If you call &error without explicitly calling unlock_file(), will locks
>>>created
>>>by lock_file be released?  I looked at the code and it was not readily
>>>apearant
>>>that this was the case, but I could have missed something.
>>>
>>
>>No, they won't .. however, because the lock files contain the PID of the
>>process that created them, as soon as it exits they will effectively be
>>
> unlocked.
> 
>>However, it is probably better for the &error function to call
>>
> &unlock_all_files,
> 
>>so that useless .lock files aren't left hanging around.
>>
> Another possibility is to create an END subroutine inside the web-lib.pl
> library
> that calls &unlock_all_files().  In this way you can ensure that if the
> script exits
> this will be called.  This won't catch signals as I recall, but you can
> also
> settup various signal handlers to call END() if you wanted to go the extra
> mile
> to insure lock files go away when they are supposed to.  This is also a
> good
> strategy for dealing with pid files and things like that.
> 
> Cheers...james

Unfortunately, miniserv already defines an END function that is uses for

a different purpose .. Is it possible to have more than one though, in different
packages?

  - Jamie

Patch for run level change ACL.

[<jam...@te...>]

This patch is against 0.990.  Unfortunately, I can't stay in step with you
Jamie, as I just have
to get our product  (which includes webmin) out the door.

Anyway,  I wanted to be able to allow someone to change run levels, but not
what started in a particular
run level, so I made it so (-:   The patch is of course below.

One thing I am not positive about is whether I should have included code to
check for:

     if ($config{'init_base'})

around the code that showed the run levels and got them.

Also, I have not added this, as I would like to question you concerning it,
but I think it would be nice
if the current runlevel was included in the init screen.   On RH if you
have the SysVinit package installed
you can use the runlevel command to determine the current run level.  This
may not be prevalant on
all linuxes.   The "who -r" command works on Solaris, but is not shared by
linux.  Anyway, how would
you implement this?  Would you have a config entry that listed the command,
or do something like the
OS depenedent libraries for the Net module?

Cheers...james


diff -urN .delta/init/acl_security.pl current/init/acl_security.pl
--- .delta/init/acl_security.pl    Tue Sep 17 15:39:25 2002
+++ current/init/acl_security.pl   Tue Sep 17 16:05:33 2002
@@ -16,6 +16,14 @@
 printf "<input type=radio name=bootup value=0 %s> $text{'no'}</td> </tr>\n",
     $_[0]->{'bootup'} ? "" : "checked";

+#
+# Change Run Level
+print "<tr> <td><b>$text{'change_runlevel'}</b></td><td>\n";
+printf "<input type=radio name=change_runlevel value=1 %s> $text{'yes'}\n",
+    $_[0]->{'change_runlevel'} ? "checked" : "";
+printf "<input type=radio name=change_runlevel value=0 %s> $text{'no'}</td> </tr>\n",
+    $_[0]->{'change_runlevel'} ? "" : "checked";
+
 print "<tr> <td><b>$text{'acl_reboot'}</b></td> <td>\n";
 printf "<input type=radio name=reboot value=1 %s> $text{'yes'}\n",
     $_[0]->{'reboot'} ? "checked" : "";
@@ -33,6 +41,7 @@
 # Parse the form for security options for the init module
 sub acl_security_save
 {
+$_[0]->{'change_runlevel'} = $in{'change_runlevel'};
 $_[0]->{'bootup'} = $in{'bootup'};
 $_[0]->{'reboot'} = $in{'reboot'};
 $_[0]->{'shutdown'} = $in{'shutdown'};
diff -urN .delta/init/change_rl.cgi current/init/change_rl.cgi
--- .delta/init/change_rl.cgi Tue Sep 17 15:39:25 2002
+++ current/init/change_rl.cgi     Tue Sep 17 16:54:44 2002
@@ -5,7 +5,7 @@
 require './init-lib.pl';
 &ReadParse();
 %access = &get_module_acl();
-$access{'bootup'} || &error($text{'change_ecannot'});
+$access{'change_runlevel'} || &error($text{'change_ecannot'});

 &header($text{'change_title'}, "");
 print "<hr>\n";
diff -urN .delta/init/index.cgi current/init/index.cgi
--- .delta/init/index.cgi     Tue Sep 17 15:39:25 2002
+++ current/init/index.cgi    Tue Sep 17 16:10:41 2002
@@ -7,6 +7,11 @@
 require './hostconfig-lib.pl';
 &header($text{'index_title'}, "", undef, 1, 1);
 %access = &get_module_acl();
+
+#
+# Determine the runlevels
+@runlevels = &list_runlevels();
+
 print "<hr>\n";

 if ($config{'hostconfig'} && $access{'bootup'}) {
@@ -45,7 +50,6 @@
          push(@actsf, $ac[0] =~ /^\// ? $ac[0]
                              : "$config{'init_dir'}/$ac[0]");
          }
-    @runlevels = &list_runlevels();
     foreach $r (@runlevels) {
          foreach $w ("S", "K") {
               foreach $a (&runlevel_actions($r, $w)) {
@@ -163,20 +167,6 @@
     print "<a href='edit_action.cgi?2'>$text{'index_add'}</a><p>\n";
     print "<hr>\n";

-    # Show runlevel switch form
-    print "<form action=change_rl.cgi>\n";
-    print "<table width=100%>\n";
-
-    print "<tr> <td nowrap><input type=submit ",
-          "value='$text{'index_rlchange'}'>\n";
-    print "<select name=level>\n";
-    foreach $r (@runlevels) {
-         printf "<option %s>%s\n",
-              $r eq $boot[0] ? "selected" : "", $r;
-         }
-    print "</select></td> <td>$text{'index_rlchangedesc'}</td> </tr>\n";
-
-    print "</table></form><hr>\n";
     }
 elsif ($config{'local_script'} && $access{'bootup'}) {
     # Display local bootup script
@@ -200,6 +190,22 @@
     print "<hr>\n";
     }

+# Show runlevel switch form
+if($access{'change_runlevel'}) {
+    print "<form action=change_rl.cgi>\n";
+    print "<table width=100%>\n";
+
+    print "<tr> <td nowrap><input type=submit ",
+          "value='$text{'index_rlchange'}'>\n";
+    print "<select name=level>\n";
+    foreach $r (@runlevels) {
+         printf "<option %s>%s\n",
+              $r eq $boot[0] ? "selected" : "", $r;
+         }
+    print "</select></td> <td>$text{'index_rlchangedesc'}</td> </tr>\n";
+    print "</table></form><hr>\n";
+    }
+
 # reboot/shutdown buttons
 print "<table cellpadding=5 width=100%>\n";
 if ($access{'reboot'}) {
diff -urN .delta/init/lang/en current/init/lang/en
--- .delta/init/lang/en  Tue Sep 17 15:39:25 2002
+++ current/init/lang/en Tue Sep 17 16:06:42 2002
@@ -25,6 +25,7 @@
 index_rlstop=Stop all actions in runlevel:
 index_rlstartdesc=Clicking this button will cause all actions that are setup to be stopped when exiting the selected runlevel to be stopped
immediately. Be careful, as this may result in the shutdown of your system.

+change_runlevel=Can Change Runlevel?

 edit_title=Edit Action
 create_title=Create Action

Re: &error() and locks...

[<jam...@te...>]

> > Hi Jamie,
> >
> > If you call &error without explicitly calling unlock_file(), will locks
> > created
> > by lock_file be released?  I looked at the code and it was not readily
> > apearant
> > that this was the case, but I could have missed something.
>
>
> No, they won't .. however, because the lock files contain the PID of the
> process that created them, as soon as it exits they will effectively be
unlocked.
>
> However, it is probably better for the &error function to call
&unlock_all_files,
> so that useless .lock files aren't left hanging around.
Another possibility is to create an END subroutine inside the web-lib.pl
library
that calls &unlock_all_files().  In this way you can ensure that if the
script exits
this will be called.  This won't catch signals as I recall, but you can
also
settup various signal handlers to call END() if you wanted to go the extra
mile
to insure lock files go away when they are supposed to.  This is also a
good
strategy for dealing with pid files and things like that.

Cheers...james

Re: New Installation

[Joe C. <jo...@sw...>]

Method of installation?

Regardless of installation method, though, something is broken on your 
system--neither rpm nor tar should segfault.  Make sure you have all of 
the errata from Red Hat installed.

Irshadh Abdul Rasheed wrote:
> Hi
> Im a new user to webmin. i was trying to install webmin on redhat 7.3
> and while trying to install its giving an error segementation fault error.
> can any one tell me what is the problem. 
> 
> 
> irshadh

-- 
Joe Cooper <jo...@sw...>
Web caching appliances and support.
http://www.swelltech.com

New Installation

[Irshadh A. R. <irs...@ya...>]

Hi
Im a new user to webmin. i was trying to install webmin on redhat 7.3
and while trying to install its giving an error segementation fault error.
can any one tell me what is the problem. 


irshadh


__________________________________________________
Do you Yahoo!?
Yahoo! News - Today's headlines
http://news.yahoo.com

Re: &error() and locks...

[Jamie C. <jca...@we...>]

jam...@te... wrote:

> Hi Jamie,
> 
> If you call &error without explicitly calling unlock_file(), will locks
> created
> by lock_file be released?  I looked at the code and it was not readily
> apearant
> that this was the case, but I could have missed something.


No, they won't .. however, because the lock files contain the PID of the
process that created them, as soon as it exits they will effectively be unlocked.

However, it is probably better for the &error function to call &unlock_all_files,
so that useless .lock files aren't left hanging around.

  - Jamie

&error() and locks...

[<jam...@te...>]

Hi Jamie,

If you call &error without explicitly calling unlock_file(), will locks
created
by lock_file be released?  I looked at the code and it was not readily
apearant
that this was the case, but I could have missed something.

Thanks...james

DHCP Leases in Subnets

[Stephens, B. {PBSG} <Bil...@pb...>]

I just pulled down the v1.0 of webmin.  The Active DHCP Leases by Subnet
made my week! Thanks!!!

 
 

Bill Stephens
PepsiCo Business Solutions Group
Bil...@pb...
972.376.7155      fax: 972.376.6019
 
 

Re: Need to limit turning on ip_forwarding...

[<jam...@te...>]

> Yeah, due to the different way each operating system and version of linux
> stores routing information and the different details that are stored,
there
> are currently 12 different versions of that function! Some like
turbo-linux-lib.pl,
> msc-linux-lib.pl, mandrake-linux-lib.pl and cobalt-linux-lib.pl are
exactly the
> same as redhat-linux-lib.pl (they are really just symlinks on my
development system).
>
> Because of the different options each OS offers, the user interface
cannot be
> the same on all of them .. which would make the abstraction that you
suggested
> tough. I know, because I tried it before :(  The DNS form on the other
hand does
> do that kind of abstraction, because every OS has basically the same
options ..

I see the light now.  I spent most of the day wrestling with the router
code to
understand it better, and now I can see really for now the best thing to do
is to
get it working in the redhat one.  The thing that was the biggest clue was
the net
versus host routes in the redhat config.  This of course makes no sense
under solaris.
Anyway, I guess I am going to get it working under redhat linux the way my
company
wants it, and if you find the changes usable then it should be easy to port
to the ones that it applies.

Actually, I suspect changes I make to the routing part even still will not
be
acceptable to you, as one of the things we want to do is not use RedHat's
GATEWAY and GATEWAYDEV entries for default routes.  The reason is that we
in
many cases may need multiple default routes, and that mechanism does not
map well.
So probably what I will end up with is something that Shows:

     - whether or not ip forwarding is turned on.
     - Static Routes
     - active routes.
     - a button beside each static route to remove it or edit it.
     - a button to add a new static route.
     - a button to alter the ip forwarding state.

I am not sure if I want to have a button to alter non-static routes.
If I did, I would certainly have that controlled by a seperate acl.
Anyway, the thing will look a lot more like the interfaces sub module,
and would store default routes along with other routes in static route
file.

So when I am done, it will be radically different.  I will send a patch
anyway,
but I am not sure you will want it (-;

Cheers...james

Re: Network Module Interface no add patch...

[<jam...@te...>]

Its based off of the .990 sources.  I was using that because it was
supposed to be the latest
stable one until today. ...james




Jamie Cameron <jca...@we...>@lists.sourceforge.net on 09/10/2002
09:18:15 PM

Please respond to web...@li...

Sent by:  web...@li...


To:   web...@li...
cc:
Subject:  Re: Network Module Interface no add patch...


That sounds like a good patch, but I had trouble applying it.
Is your code based off the 0.999 version of webmin? Some similar
access control options were contributed by another user since
the 0.990 release ..

  - Jamie

jam...@te... wrote:

> Hi Jamie,
>
> It was desirable in our deployment of webmin to have the option the
> interfaces submodule of the
> network module, to make it so that a user could could view and edit an
> interface but they
> could not create a new one.  At the end of this email I have attached a
> patch that adds this
> ability.  The patch includes the following changes (high level):
>
>      - Adds the option to add an interface (yes/no) to the
acl_security.pl
> file.
>      - Adds text for not being able to add an interface to lang/en.
>      - Adds text for not being able to view an interface.
>      -  list_ifcs.cgi now uses the cannot view interface error message if
> they
>          cannot even view the interface.
>      - list_ifcs.cgi does not show a link to add an interface if they
user
> does not have
>        that right.
>      - edit_aifcs.cgi checks to see how it was called.  If it was called
to
> create a
>         an interface it determines if the user can and if not gives an
> error.  If it
>         was called to edit an interface it checks if the user can edit
it.
> If he/she can't
>         then it gives an error.
>
> I set it up so that even if a user could not edit an interface, if
allowed
> to he/she can add
> one.  I actually not totally sure of that decision; it seemed like the
> flexible thing to do, but
> it would have been a tiny bit less code changes if I would have added
> append to edit, view, no.
>
> I did not update all the language files (only english).  I could have
done
> the Espanol one, but
> I was a little unsure of the exact grammer.
>
> Anyway, I hope the patch will be useful.
>
> Cheers...james





-------------------------------------------------------
In remembrance
www.osdn.com/911/
-
Forwarded by the Webmin development list at web...@we...
To remove yourself from this list, go to
http://lists.sourceforge.net/lists/listinfo/webadmin-devel

Re: Need to limit turning on ip_forwarding...

[Jamie C. <jca...@we...>]

jam...@te... wrote:

> Hi Jamie,
> 
> I need to be able to limit who can turn on and off ip forwarding (Act as
> router? question in
> the routing and gateways sub module of the net module) while still allowing
> them to
> add static routes.  I thought it would be a trivial change but then I found
> that the manual
> polymorphism of your net-lib.pl library (-;
> 
> It looks like the routing_input() function is what is responsible for
> painting the screen with
> the form that:
> 
>      - shows all static routes.
>      - allows the selection of "Act as router?"/ip_forwarding.
> 
> The problem is that this funciton lives in what looks like 14 different
> libraries each representing
> the way to do things on one specific platform.   So if I fix it in the
> library redhat-linux-lib.pl, the fix still
> needs to be propagated to 13 other platforms, most of which I cannot test.
> That being said,
> what I need to control really should operate from a forms perspective
> exactly the same no
> matter what platform I am on.   What it really looks like needs to be done
> is to architect things
> just a little different so that you have a seperate output routine that is
> common to all platforms, and then
> have each platform have instead a function that gathers the information
> into a "standard" hash, and
> pass that back to the format routine....sigh.
> 
> This appears to be the right thing to do, but runs the risk of even more
> problems at first, and would
> call for even better testing on the other platforms.  What do you think
> Jamie.  I am willing to reconstruct
> it as mentioned above, but is this a desirable thing for you and do you
> have the systems and or
> time to test it?  I could just make it work the way I want under redhat
> linux but it seems to be a desirable
> feature for others?  What seems amicable to you?


Yeah, due to the different way each operating system and version of linux
stores routing information and the different details that are stored, there
are currently 12 different versions of that function! Some like turbo-linux-lib.pl,
msc-linux-lib.pl, mandrake-linux-lib.pl and cobalt-linux-lib.pl are exactly the
same as redhat-linux-lib.pl (they are really just symlinks on my development system).

Because of the different options each OS offers, the user interface cannot be
the same on all of them .. which would make the abstraction that you suggested
tough. I know, because I tried it before :(  The DNS form on the other hand does
do that kind of abstraction, because every OS has basically the same options ..

  - Jamie

Re: net module list vs edit question...

[Jamie C. <jca...@we...>]

jam...@te... wrote:

> Hi Jamie,
> 
> When you go to view some of the sub modules in the net module, though they
> are listed as
> just list CGI's ( list_routes.cgi for instance), they default to allowing
> you to edit them (they are in
> a form).  How opposed would you be to me patching them such that instead
> you seem them
> in text (not in a form) and there is an edit button there if you are
> allowed to edit them?
> 
> Of course I am asking because that is how I wish it to work in order to
> minimize the risk of
> a customer doing the wrong thing (granted you still have to click save).
> Still, if I do this I want
> it to work in a way that is acceptable to you.


Maybe a better option would be to have the routes displayed as text if
the user is not allowed to edit them, or input fields if he is? I guess having
a table of routes would be nice if there was a very large number of them, but
that is not usually the case ..

  - Jamie

Re: Network Module Hosts no add patch...

[Jamie C. <jca...@we...>]

jam...@te... wrote:

> This patch allows one to allow editing of hosts names but not allow adding
> or vice versa/
> This patch works the same thing the one I just sent for the interfaces.


This one didn't apply either, probably for the same reason ..

  - Jamie

Re: Network Module Interface no add patch...

[Jamie C. <jca...@we...>]

That sounds like a good patch, but I had trouble applying it.
Is your code based off the 0.999 version of webmin? Some similar
access control options were contributed by another user since
the 0.990 release ..

  - Jamie

jam...@te... wrote:

> Hi Jamie,
> 
> It was desirable in our deployment of webmin to have the option the
> interfaces submodule of the
> network module, to make it so that a user could could view and edit an
> interface but they
> could not create a new one.  At the end of this email I have attached a
> patch that adds this
> ability.  The patch includes the following changes (high level):
> 
>      - Adds the option to add an interface (yes/no) to the acl_security.pl
> file.
>      - Adds text for not being able to add an interface to lang/en.
>      - Adds text for not being able to view an interface.
>      -  list_ifcs.cgi now uses the cannot view interface error message if
> they
>          cannot even view the interface.
>      - list_ifcs.cgi does not show a link to add an interface if they user
> does not have
>        that right.
>      - edit_aifcs.cgi checks to see how it was called.  If it was called to
> create a
>         an interface it determines if the user can and if not gives an
> error.  If it
>         was called to edit an interface it checks if the user can edit it.
> If he/she can't
>         then it gives an error.
> 
> I set it up so that even if a user could not edit an interface, if allowed
> to he/she can add
> one.  I actually not totally sure of that decision; it seemed like the
> flexible thing to do, but
> it would have been a tiny bit less code changes if I would have added
> append to edit, view, no.
> 
> I did not update all the language files (only english).  I could have done
> the Espanol one, but
> I was a little unsure of the exact grammer.
> 
> Anyway, I hope the patch will be useful.
> 
> Cheers...james

Re: [BUG-REPORT] DHCP config on RH 7.x

[Jamie C. <jca...@we...>]

Steve M Bibayoff wrote:

> I don't know where Jamie(or anybody else) cares to have bug reports
> submited, and noticing that the webmin bug tracking system seems to be
> rarely used recently, I'm submitting it here first. Sorry if it was
> preferred to have been submitted somewhere else first, will do next time.
> 
> Under Servers->DHCPD->Module_Config if you try to change the "Interfaces
> to run DHCP on" from "Automatic" to a user defined (say eth1, or
> anything besides eth0), it doesn't register the change. Under RH
> 7.[1,2,3], the file /etc/sysconfig/dhcpd needs to be ammended with eth1
> (or whatever the interface(s) needed to be used) after "DHCPDARGS=" .
> 
> I don't know if this is required before <RH7.0 or any other  distro. And
> I currently don't have time to figure out what is needed to fix it(I
> currently just edit by hand). If nobody submits a fix by wedsnday, I
> would have time then to dive in and see what needs changing.


Yeah, that option is rather misleading as it only applies when webmin starts
dhcpd directly, not through redhat's init script. What really needs to be done
is to take it out of the module config page and put it on one of the other forms.
The code could then update the /etc/sysconfig/dhcpd file so that redhat's init
scripts use the right interface, as well as the module config so that when webmin
starts dhcpd directly it adds the right interface to the command as well.

  - Jamie