net-snmp-coders — Discussion of internal project coding (Please use this OR -users, NOT both!)

Re: enterprises.ucdavis.systemStats.ssCpuUser & HP-UX 10.20

[Stan B. <st...@aw...>]

On Wed Nov  8 12:48:26 2000 Dave Shield wrote...
>
>
>  The hardest part of the job will probably be in obtaining the information
>from the kernel - in particular, working out where to look in the first place!
>A lot of these interfaces are not documented very well (if at all), and in
>the worst case, you may need to delve directly into the running kernel.
>Try and avoid that if you can, but it's not always possible.

	Actually, once I know what I need hee, that should be simple. I am in
	the fortunate postion of having a current support contract with HP, and
	in general their support people are very good. So if I can properly
	phrase teh question, then I should be in good shape.

>
>  Another useful source of information is if you can find another
>(open source) tool that reports on the same sort of thing.  The code for
>that can often be helpful - no point in re-inventing the wheel....
>
>
	For FreeBSD, top, and systat should provide details of what I need.
	Probably OpenBSD also.

	Thanks for the help.


-- 
Stan Brown     st...@aw...                                    843-745-3154
Charleston SC.
-- 
Windows 98: n.
	useless extension to a minor patch release for 32-bit extensions and
	a graphical shell for a 16-bit patch to an 8-bit operating system
	originally coded for a 4-bit microprocessor, written by a 2-bit 
	company that can't stand for 1 bit of competition.
-
(c) 2000 Stan Brown.  Redistribution via the Microsoft Network is prohibited.

Re: snmpd dies when agentx subagent dies

[Lawrence G. <le...@an...>]

   Cc: net...@li...
   Date: Wed, 08 Nov 2000 09:42:11 +0000
   From: Dave Shield <D.T...@cs...>

   > I've been developing an application with an AgentX subagent for
   > monitoring purposes.
   > 
   > I'm using the latest code out of cvs on a Redhat6-derived system, and
   > I've noticed that when my application dies (if I kill it, or it exits,
   > or whatever) and I run an snmp query against that part of the tree,
   > snmpd will die.

   When you shutdown your subagent, do you do this cleanly?
   In particular, does the subagent unregister from the master agent
   before exiting?

Nope; I wasn't even aware that there was a call for this.  I'll make
that change (this process already catches signals, so it should be
extremely easy).

     Which version of the code are you using?   4.1.2 or the current
   development code (i.e. 4.2.pre1 or the CVS tree) ?
   The current development code includes support to detect a dead subagent,
   and unregister it anyway - but I'm not sure how well tested this is.
   The released code does not include this support, so is reliant on
   subagents being "well behaved".

I'm using the CVS tree, so it appears that the code dealing with badly
behaved subagents isn't working for me properly.

Larry

Re: enterprises.ucdavis.systemStats.ssCpuUser & HP-UX 10.20

[Dave S. <D.T...@cs...>]

> 	OK, it looks like, if I want support for HP-UX, and FreebSD in this MIB
> 	(and I do). I will have to get my hands dirty :-)

That's how this project works - you want it, you write it!  :-)
I'm probably unusual, in that I spend all my time writing code,
and never get around to doing anything with it....


> 	I'm not the worlds greatest kernel hacker, but I doo speak passible C.
> 	So if someone can give me a frw pointers on how to approach making this
> 	MIB support all of it's functionality under a new OS. I will start
> 	checking inot HP-UX 10.20m and FreeBSD 3 & 4. Perhaps OpenBSD, also.

Well, for the CPU statistics, your best approach is probably to start with
one of the equivalent files for one of the other architectures, make a copy
of that and change it to match the HP-UX way of working.
  If you need information about how everything fits together, try reading
through
		http://www.csc.liv.ac.uk/~daves/Misc/UCD

  The hardest part of the job will probably be in obtaining the information
from the kernel - in particular, working out where to look in the first place!
A lot of these interfaces are not documented very well (if at all), and in
the worst case, you may need to delve directly into the running kernel.
Try and avoid that if you can, but it's not always possible.

  Another useful source of information is if you can find another
(open source) tool that reports on the same sort of thing.  The code for
that can often be helpful - no point in re-inventing the wheel....

Have Fun.

Dave

Re: diskio MIB

[Stan B. <st...@aw...>]

On Tue Nov  7 16:04:36 2000 =?iso-8859-1?Q?Ragnar_Kj=F8rstad?= wrote...
>
>On Tue, Nov 07, 2000 at 01:44:04PM -0500, Stan Brown wrote:
>> >Stan> Now, do I need to add anything to snmpd.conf to enable this? If I do a
>> >Stan> snmpwalk of the whole ucd-snmp MIB, will the disk I/O data show
>> >Stan> up?
>> >
>> >It should if you ran configure as mentioned above.
>> 
>> 	Yes, I am getting something, now if only I can figure out how to decode
>> 	it :-( Seems to be some sort of table?
>
>
>Yes, this is a sample from one of our servers:
>
>So to find the number of bytes written to sd1, you look at
>.....diskIONWritten.3 (because index 3 is sd1)
>

	Thans to your kind help, I thought I had this under control, but alas,
	I don't Let me show you what I am getting on a Solaris box:


snmpwalk localhost public .1.3.6.1.4.1.2021 | grep diskIOTable 
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.diskIOEntry = 1
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.2 = 2
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.3 = 3
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.4 = 4
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.5 = 5
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.6 = 6
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.7 = 7
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.8 = 8
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.9 = 9
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.10 = 10
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.11 = 11
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.12 = 12
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.13 = 13
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.14 = 14
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.15 = 15
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.16 = 16
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.17 = 17
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.18 = 18
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.19 = 19
enterprises.ucdavis.ucdExperimental.ucdDiskIOMIB.diskIOTable.20 = 20

	Is it possible I still don't have something configured corectly on this
	machine?

-- 
Stan Brown     st...@aw...                                    843-745-3154
Charleston SC.
-- 
Windows 98: n.
	useless extension to a minor patch release for 32-bit extensions and
	a graphical shell for a 16-bit patch to an 8-bit operating system
	originally coded for a 4-bit microprocessor, written by a 2-bit 
	company that can't stand for 1 bit of competition.
-
(c) 2000 Stan Brown.  Redistribution via the Microsoft Network is prohibited.

Re: enterprises.ucdavis.systemStats.ssCpuUser & HP-UX 10.20

[Wes H. <wjh...@uc...>]

>>>>> On Wed, 8 Nov 2000 09:21:05 -0500 (EST), "Stan Brown" <st...@aw...> said:

Stan> Is there a problem obtaining these values under HP-UX, or is it
Stan> just something that has not been done?

Those aren't supported for HPUX (yet), sorry.

-- 
Wes Hardaker
Please mail all replies to net...@li...

Re: Need Help

[Wes H. <wjh...@uc...>]

>>>>> On Wed, 08 Nov 2000 11:12:24 +1300, "Simon Li" <Sim...@te...> said:

Simon> I would like to free all memory allocated by init_snmp and
Simon> snmp_sess_init, I try to use snmp_shutdown to free, but I can
Simon> not free all memory. Please tell me if you know it.

You might try the latest pre-release (get it from
http://sourceforge.net/projects/net-snmp), which should come closer to
freeing everything.

-- 
Wes Hardaker
Please mail all replies to net...@li...

Re: enterprises.ucdavis.systemStats.ssCpuUser & HP-UX 10.20

[Stan B. <st...@aw...>]

On Wed Nov  8 10:24:45 2000 Dave Shield wrote...
>
>
>> 	Is there a problem obtaining these values under HP-UX, or is it just
>> 	something that has not been done?
>
>Dunno - it's clearly not been done, but without actually trying it,
>I'm not sure how difficult it would be.
>  And no - I'm *not* volunteering!  :-)

	OK, it looks like, if I want support for HP-UX, and FreebSD in this MIB
	(and I do). I will have to get my hands dirty :-)

	I'm not the worlds greatest kernel hacker, but I doo speak passible C.
	So if someone can give me a frw pointers on how to approach making this
	MIB support all of it's functionality under a new OS. I will start
	checking inot HP-UX 10.20m and FreeBSD 3 & 4. Perhaps OpenBSD, also.

-- 
Stan Brown     st...@aw...                                    843-745-3154
Charleston SC.
-- 
Windows 98: n.
	useless extension to a minor patch release for 32-bit extensions and
	a graphical shell for a 16-bit patch to an 8-bit operating system
	originally coded for a 4-bit microprocessor, written by a 2-bit 
	company that can't stand for 1 bit of competition.
-
(c) 2000 Stan Brown.  Redistribution via the Microsoft Network is prohibited.

Re: UID GID patch (2)

[Wes H. <wjh...@uc...>]

>>>>> On Tue, 07 Nov 2000 16:06:18 -0800, Harrie Hazewinkel <ha...@co...> said:

Harrie> I also looked into the file-configuration code myself as well.
Harrie> SOme parts look a bit complex.

It is somewhat complex, admittedly.  Most people, though, shouldn't
have to deal with the complex parts.

Harrie> Also the help-field is never really used.

Yes it is.  It's used to print out the -H help list (which is a list
of tokens that a given command supports).

Harrie> An idea would be to have all token/line handling functions
Harrie> return a 'char *'. In case it is successfull it return 'NULL'
Harrie> in case of preblems "string with message". This would reduce
Harrie> the token mechanism I think.

Thats another way to do it.  config_perror() does roughly the same
thing, but allows you to specify whether something is an error or a
warning which a char * return wouldn't let you do.
-- 
Wes Hardaker
Please mail all replies to net...@li...

Re: Pluggable transport domains via register_readfd-like API?

[John N. <jb...@ca...>]

> Yes - that's more-or-less what I had in mind.  And similarly, old and new
> versions of the utility routines 'generic_header' and the like.
> 
>   I'd also want to aim for an extensible API, so that further changes
> to the interface would be source-code compatible.   This probably
> means using some sort of structure as (one of) the parameters, which
> could be extended without affecting the rest of the code.

  Mmm, good point.  Of course struct snmp_session and struct snmp_pdu are 
interesting candidates for (two of) the parameters, containing as they do 
lots of juicy "administrative" information, as you put it...

>   (I've even been toying with the idea of using the existing 'struct variable'
> parameter for this, hence avoiding the need to change the API at all.  But
> I'm not convinced that's necessary a sensible way to go....)

  Me neither... ;-)

>   I'd been more thinking about the next stage - how to interpret the
> data read from the socket, and how to encode data before sending it
> out.  Part of this has been done for the AgentX work (which is totally
> external to the library) but there's still a lot of special-casing
> within the library.   I'd like to pull out the SNMPv1/2c code into
> one file, and the SNMPv3 code into another - both separate from the
> driving code.
> 
>    This is distinct from (but compatible with) your suggestion, and again
> should have the effect of slimming down snmp_api.c and (hopefully) making
> it easier to follow.

  I see what you are saying -- in fact it did strike me there was a lot of 
v3 specific stuff in there that could move.  Of course if that moved it 
would only be right to move out the v1/2c stuff too.  Making both sets of 
changes would actually make the internal structure of the code look a lot 
like the abstract model used in RFC2571 and RFC2572, which has a certain 
logical appeal.  (It may also have the useful side-effect of making the 
learning curve shallower for people new to the toolkit).

>   You see what I mean when I talked about rebuilding things from the
> ground up? :-)

  Yup!

>   Any other ideas anyone?

  Yeah, c'mon, let's hear them.

John

-- 
// Dr. John Naylon // Senior Software Engineer // Cambridge Broadband Ltd.
// The future of broadband wireless access  // www.cambridgebroadband.com

Re: enterprises.ucdavis.systemStats.ssCpuUser & HP-UX 10.20

[Dave S. <D.T...@cs...>]

> 	I can't seem to get smpwalk to return values for
> 	enterprises.ucdavis.systemStats.ssCpuUser &
> 	enterprises.ucdavis.systemStats.ssCpuSystem
> 	from my HP-UX machines, runing the agaent from teh latest CVS sources.

Not surprising - really:

$ grep vmstat agent/mibgroup/ucd_snmp.h 
config_arch_require(solaris2, ucd-snmp/vmstat_solaris2)
config_arch_require(linux, ucd-snmp/vmstat)
config_arch_require(freebsd2, ucd-snmp/vmstat_freebsd2)
config_arch_require(freebsd3, ucd-snmp/vmstat_freebsd2)
config_arch_require(freebsd4, ucd-snmp/vmstat_freebsd2)
config_arch_require(netbsd1, ucd-snmp/vmstat_netbsd1)
config_arch_require(bsdi4, ucd-snmp/vmstat_bsdi4)


This particular module is not supported under HP-UX.
(Use the Source, Stan! :-)  )

> 	Is there a problem obtaining these values under HP-UX, or is it just
> 	something that has not been done?

Dunno - it's clearly not been done, but without actually trying it,
I'm not sure how difficult it would be.
  And no - I'm *not* volunteering!  :-)

Dave

enterprises.ucdavis.systemStats.ssCpuUser & HP-UX 10.20

[Stan B. <st...@aw...>]

	I can't seem to get smpwalk to return values for
	enterprises.ucdavis.systemStats.ssCpuUser &
	enterprises.ucdavis.systemStats.ssCpuSystem
	from my HP-UX machines, runing the agaent from teh latest CVS sources.

	Is there a problem obtaining these values under HP-UX, or is it just
	something that has not been done?



-- 
Stan Brown     st...@aw...                                    843-745-3154
Charleston SC.
-- 
Windows 98: n.
	useless extension to a minor patch release for 32-bit extensions and
	a graphical shell for a 16-bit patch to an 8-bit operating system
	originally coded for a 4-bit microprocessor, written by a 2-bit 
	company that can't stand for 1 bit of competition.
-
(c) 2000 Stan Brown.  Redistribution via the Microsoft Network is prohibited.

Re: Pluggable transport domains via register_readfd-like API?

[Dave S. <D.T...@cs...>]

>         One possibility might be to have a variant of register_mib 
> that lets you specify which callback API you want for your var_ and write_ 
> routines -- the Classic or Advanced API.  Then you could migrate code on a 
> per-MIB-module (or finer) granularity to the new API.

Yes - that's more-or-less what I had in mind.  And similarly, old and new
versions of the utility routines 'generic_header' and the like.

  I'd also want to aim for an extensible API, so that further changes
to the interface would be source-code compatible.   This probably
means using some sort of structure as (one of) the parameters, which
could be extended without affecting the rest of the code.
  (I've even been toying with the idea of using the existing 'struct variable'
parameter for this, hence avoiding the need to change the API at all.  But
I'm not convinced that's necessary a sensible way to go....)



>   I've been looking at snmp_api.c more closely today -- I see what you
> mean when you said it made your head hurt now!

<grin>

>  I was thinking of something vaguely like this:
> 
>   New header snmp_transport.h defines a struct snmp_transport which
> provides a transport-independent API

  Interesting idea.   I hadn't been thinking along _quite_ those lines,
but that sounds sensible.

  I'd been more thinking about the next stage - how to interpret the
data read from the socket, and how to encode data before sending it
out.  Part of this has been done for the AgentX work (which is totally
external to the library) but there's still a lot of special-casing
within the library.   I'd like to pull out the SNMPv1/2c code into
one file, and the SNMPv3 code into another - both separate from the
driving code.

   This is distinct from (but compatible with) your suggestion, and again
should have the effect of slimming down snmp_api.c and (hopefully) making
it easier to follow.

  You see what I mean when I talked about rebuilding things from the
ground up? :-)

  Any other ideas anyone?

Dave

4.2.pre1 dumps core on Solaris 2.6

[RADMAN S. <Ste...@CT...>]

Here are my results from a build of ucd-snmp-4.2.pre1 on Solaris 2.6:

Environment:
	SunOS idc009 5.6 Generic_105181-17 sun4u sparc SUNW,Ultra-1
	gcc version 2.95.2 19991024 (release)

Config:
	modules="host smux target notification agentx ucd-snmp/diskio
ucd-snmp/dlmod"
	./configure --prefix=$PREFIX \
	        --with-defaults \
	        --with-sys-contact="unknown" \
	        --with-sys-location="unknown" \
	        --with-mib-modules="$modules" \
	        --with-openssl=/opt/IDCsoh \
	        --without-root-access

Results:
- compilation went fine
- tests all failed with coredump (Segmentation Fault or Bus Error).
- same result when I configure with defaults
- same result when I run the tests as root
- "MIBS= ./apps/snmpwalk otherhost community succeeds (showing only
numerical OIDs)
- "MIBDIRS=./mibs ./apps/snmpwalk otherhost community" dumps core
- "MIBDIRS=./mibs ./apps/snmptranslate system" dumps core

Guess:
The problem seems to be related to MIB (file) loading.

typescript of build is attached.
 <<typescript>> 

Debug:
IDCsoh@idc009:work/ucd-snmp-4.2.pre1/apps/.libs$ ./snmptranslate system
Segmentation Fault(coredump)
IDCsoh@idc009:work/ucd-snmp-4.2.pre1/apps/.libs$ gdb ./snmptranslate core
GNU gdb 4.18
Copyright 1998 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "sparc-sun-solaris2"...(no debugging symbols
found)...
Core was generated by `./snmptranslate system'.
Program terminated with signal 11, Segmentation Fault.
Reading symbols from
/ctbto/idc/infra/ho/source/build/IDCsoh/work/ucd-snmp-4.2.pre1/snmplib/.libs
/libsnmp-0.4.2.0.1.so..
.(no debugging symbols found)...done.
Reading symbols from /usr/lib/libdl.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libkvm.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libkstat.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libm.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libelf.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libnsl.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libsocket.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libc.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libmp.so.2...(no debugging symbols
found)...done.
Reading symbols from /usr/platform/SUNW,Ultra-1/lib/libc_psr.so.1...(no
debugging symbols found)...done.
Reading symbols from /usr/lib/nss_nisplus.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/libdoor.so.1...(no debugging symbols
found)...done.
Reading symbols from /usr/lib/nss_files.so.1...(no debugging symbols
found)...done.
---Type <return> to continue, or q <return> to quit---
#0  0xef4a4734 in strlen () from /usr/lib/libc.so.1
(gdb)
(gdb) run system
Starting program:
/ctbto/idc/infra/ho/source/build/IDCsoh/work/ucd-snmp-4.2.pre1/apps/.libs/./
snmptranslate system
(no debugging symbols found)...(no debugging symbols found)...(no debugging
symbols found)...
(no debugging symbols found)...(no debugging symbols found)...(no debugging
symbols found)...
(no debugging symbols found)...(no debugging symbols found)...(no debugging
symbols found)...
(no debugging symbols found)...(no debugging symbols found)...(no debugging
symbols found)...
(no debugging symbols found)...(no debugging symbols found)...
Program received signal SIGSEGV, Segmentation fault.
0xef4a4734 in strlen () from /usr/lib/libc.so.1
(gdb)

Sorry wasnt able to produce more in the hurry ...

Stefan Radman
Network Specialist
CTBTO/IDC

Email:	mailto://Ste...@ct...
Phone:	+43 (1) 26030-6182
Fax:	+43 (1) 260308-6182

Re: Problems building latest Perl/SNMP

[<ni...@ba...>]

On Tue, 7 Nov 2000 16:42:15 -0800 (PST) Terminator rAT <kar...@el...>
wrote:
>I just grabbed a copy of the Net-SNMP sources via anoncvs from sourceforge
>and tried to build the Perl/SNMP part on my Solaris 2.7 box.  Here is the
>build error and systems info:
>
>rat@sandbox ~/work/Net-SNMP-cvs/net-snmp/perl/SNMP 602% make
>gcc -c -I/usr/local/include -DDEBUGGING -O -DVERSION=\"3.1.0\"
-DXS_VERSION=\"3.1.0\" -fPIC
-I/usr/local/lib/perl5/5.6.0/sun4-solaris/CORE  SNMP.c

This indicates that the perl installation expects that you have done a make
install in the main directory first. I rarely use perl, so I never noticed, but
I think this should be modified to use the non-installed directories.

/Niels

-- 
Get your firstname@lastname email for FREE at http://Nameplanet.com/?su

udc-snmp-4.1.1-x86-win32 problem

[hbu <hb...@ne...>]

Hello community,

Have a binary ucd-snmp-4.1.1-win32 problem regarding mibs.

running snmpd.exe gives a error

Cannot find module <IP-MIB>: at line 0 in <none>
Cannot find module <IF-MIB>: at line 0 in <none>
Cannot find module <TCP-MIB>: at line 0 in <none>
Cannot find module <UDP-MIB>: at line 0 in <none>
Cannot find module <SNMPv2-MIB>: at line 0 in <none>
Cannot find module <SNMPv2-SNI>: at line 0 in <none>

running snmpd -M \mibs gives "invalid option: -M"
The man pages do not give an solution to this error.

Please, are you able to help me.


Kind regards,
Herman 



____________________________________________________________________
Get your own FREE, personal Netscape WebMail account today at http://home.netscape.com/webmail

Re: snmpd dies when agentx subagent dies

[Dave S. <D.T...@cs...>]

> I've been developing an application with an AgentX subagent for
> monitoring purposes.
> 
> I'm using the latest code out of cvs on a Redhat6-derived system, and
> I've noticed that when my application dies (if I kill it, or it exits,
> or whatever) and I run an snmp query against that part of the tree,
> snmpd will die.

When you shutdown your subagent, do you do this cleanly?
In particular, does the subagent unregister from the master agent
before exiting?

  Which version of the code are you using?   4.1.2 or the current
development code (i.e. 4.2.pre1 or the CVS tree) ?
The current development code includes support to detect a dead subagent,
and unregister it anyway - but I'm not sure how well tested this is.
The released code does not include this support, so is reliant on
subagents being "well behaved".

  If you're not using the development code, you should try that.
If you are, then please investigate further, and let us know how
you get on.

Dave

lm_sensors hack

[Martin O. <m...@ma...>]

I've hacked some support for Linux's lm_sensors into snmpd. I hesitate 
to call it a patch because my knowledge of MIBs and whatnot is
essentially non-existent.

Ideally I'd like two things:

1. To get some feedback and improvements from people who actually
   understand SNMP and have experience in dealing with it.

2. To get this code folded into the main distribution.

Please can someone suggest how I should go about this ?

Cheers,
-- 
Martin Oldfield,
AdamsNames Ltd.

Problems building latest Perl/SNMP

[Terminator r. <kar...@el...>]

Hi, folks.

I just grabbed a copy of the Net-SNMP sources via anoncvs from sourceforge
and tried to build the Perl/SNMP part on my Solaris 2.7 box.  Here is the
build error and systems info:

rat@sandbox ~/work/Net-SNMP-cvs/net-snmp/perl/SNMP 602% make
gcc -c -I/usr/local/include -DDEBUGGING -O -DVERSION=\"3.1.0\" -DXS_VERSION=\"3.1.0\" -fPIC -I/usr/local/lib/perl5/5.6.0/sun4-solaris/CORE  SNMP.c
SNMP.xs: In function `__translate_appl_type':
SNMP.xs:286: `TYPE_NOTIFTYPE' undeclared (first use in this function)
SNMP.xs:286: (Each undeclared identifier is reported only once
SNMP.xs:286: for each function it appears in.)
SNMP.xs: In function `__get_type_str':
SNMP.xs:540: `TYPE_NOTIFTYPE' undeclared (first use in this function)
SNMP.xs: In function `XS_SNMP__MIB__NODE_FETCH':
SNMP.xs:3145: structure has no member named `varbinds'
SNMP.xs:3145: dereferencing pointer to incomplete type
SNMP.xs:3146: dereferencing pointer to incomplete type
SNMP.xs:3146: dereferencing pointer to incomplete type
make: *** [SNMP.o] Error 1

rat@sandbox ~/work/Net-SNMP-cvs/net-snmp/perl/SNMP 603% uname -a
SunOS sandbox 5.7 Generic_106541-11 sun4u sparc SUNW,Ultra-60

rat@sandbox ~/work/Net-SNMP-cvs/net-snmp/perl/SNMP 604% cvs stat SNMP.xs 
===================================================================
File: SNMP.xs           Status: Up-to-date

   Working revision:    1.20
   Repository revision: 1.20 /cvsroot/net-snmp/net-snmp/perl/SNMP/SNMP.xs,v
   Sticky Tag:          (none)
   Sticky Date:         (none)
   Sticky Options:      (none)

Thought you'd like to know.

	-rAT

-- 
O----O        The Terminator rAT -+- ksc...@el... -+- (360)816-3837
 \Oo/    
==\/==    "We have enough youth.  How about we find the fountain of SMART?!"

Re: UID GID patch (2)

[Harrie H. <ha...@co...>]

Wes Hardaker wrote:
> 
> Harrie> Maybe the patch is not requried anymore, but just for
> Harrie> reference.
> 
> What we did was fairly similar.  I've actually changed pieces of mine
> to match yours more.
> 
> I'll still have to implement an alternative to exit() in the libsnmp.a
> code however.

I also looked into the file-configuration code myself as well.
SOme  parts look a bit complex. Also the help-field is never really
used. An idea would be to have all token/line handling functions
return a 'char *'. In case it is successfull it return 'NULL'
in case of preblems "string with message". This would reduce
the token mechanism I think.


Harrie 
0- Harrie Hazewinkel ---------------------------------------0
 mailto:ha...@co...            phone:+1-415-536-5221
0-----------------------------------------------------------0

Re: about ucd-snmp

[Wes H. <wjh...@uc...>]

>>>>> On Tue, 07 Nov 2000 11:13:43 -0800, Shiow Jen Jin <sj...@cl...> said:

Shiow> If I ignore the error message, please tell me how can I convert
Shiow> EtherLike-MIB.txt to transmission.c and transmission.h template
Shiow> by mib2c.

Didn't it generate one anyway?

-- 
Wes Hardaker
Please mail all replies to net...@li...

Re: UID GID patch (2)

[Wes H. <wjh...@uc...>]

Harrie> Maybe the patch is not requried anymore, but just for
Harrie> reference.

What we did was fairly similar.  I've actually changed pieces of mine
to match yours more.

I'll still have to implement an alternative to exit() in the libsnmp.a 
code however.

-- 
Wes Hardaker
Please mail all replies to net...@li...

Need Help

[Simon L. <Sim...@te...>]

Hi, Everyone works for snmp,
I would like to free all memory allocated by init_snmp and snmp_sess_init, I try to use snmp_shutdown to free, but I can not free all memory. Please tell me if you know it.
Thanks in advance!

Regards,
Simon

Re: UID GID patch (2)

[Harrie H. <ha...@co...>]

Wes Hardaker wrote:
> 
> >>>>> On Tue, 07 Nov 2000 10:07:51 -0800, Harrie Hazewinkel <ha...@co...> said:
> 
> Harrie> I have attached a patch that enables the configuration of the
> Harrie> uid and gid for the snmpd via the configuration file.
> 
> Actually, you didn't attach the patch :-)

Oops.

> I actually applied your earlier patch already and made similar
> modifications to the one that you did.

Maybe the patch is not requried anymore, but just for reference.


Harrie 
0- Harrie Hazewinkel ---------------------------------------0
 mailto:ha...@co...            phone:+1-415-536-5221
0-----------------------------------------------------------0

Re: Trouble Loading 4.2pre1 win32.dsw file

[Michael J. S. <msl...@is...>]

"John C. Westmoreland" wrote:
> 
> NET-SNMP-CODERS,
> 
> Has anyone had trouble loading the win32.dsw file (project file
> needed to do a Win32 build in batch mode)?
> 
> When I load it - it does nothing.  MSVC++ 6.0 just seems to ignore
> the file.
> 
> Regards,
> John


Except for the .opt and .ncb files, the other files in the Win32
subdirectory must use CR-LF (ie., PC-DOS) line termination.

Otherwise, DevStudio ignores them.

Suggest you re-unpack the tar ball using a ZIP file reader.

Best Regards,
-Mike Slifcak, Internet Security Systems, Inc.
   "The Power to Protect"

snmpd dies when agentx subagent dies

[Lawrence G. <le...@an...>]

I've been developing an application with an AgentX subagent for
monitoring purposes.

I'm using the latest code out of cvs on a Redhat6-derived system, and
I've noticed that when my application dies (if I kill it, or it exits,
or whatever) and I run an snmp query against that part of the tree,
snmpd will die.

Further, I can't rerun my subagent and have things start working
again.

Is this a known bug?  Should I do more digging?

Thanks,
Larry