The following is a series of messages I've posted to the ucd-snmp mailing list at ucdavis. I'm going to copy them here, because I'm getting a feeling that this is more of a Linux networking issue than a problem with SNMP, it's just that SNMP triggered me to the problem.. So, be warned, this message is very long, but this is also something I've been fighting with for almost two years, and nobody seems to have an answer, or even a HINT of an answer, so here is one of my last desperate acts to figure this problem out... There are two message texts contained herein... In the second message is my reply to someone who asked me a number of questions in an effort to try to diagnose the problem... As I said in the first message, I'm almost ready to give up in frustration and disgust... WinDoze should NOT work better than Linux... LOL!
Thanks in advance,
Richard W.
---
Greetings all:
I'm about at the end of my rope with this problem and am ready to throw up my hands in surrender. I've been fighting with this problem for almost 2 years, and thus far no body has been able to come up with an explanation as to WHY I'm having the problem I'm having... So... I'm hoping this list's scope is enough to find someone (or several someones) who have either seen this problem before, or have a resonable idea of what the problem is... I will include as much information as I can about the systems involved in the hopes that it will do some good. So, after that lengthy preamble, here's my problem:
I have an APC1400RMXL UPS with an SNMP management card in it. I would dearly like to be able to poll this card with the ucd SNMP package to find out the status of the UPS, and to control the UPS via remote management... I can poll the UPS on a WinDoze machine on my network, but every single Linux machine I've tried fails miserably... I've tried just about everything I can think of... I've tried machines on the "inside" of my network, I've tried my gateway machines, all manner of things... I've tried various versions of SNMPD from 3.x on up to the current 4.1.2. It doesn't seem to matter what version of Linux I use... I've tried Redhat 4.1, 5.2, and 6.0, Slackware 4.0 and 7.0, and nothing works... It seems to be the only machine that can poll this damnable UPS is a Windows 98 machine (actually 2 of them). So, here's my network topography in the hopes it might answer some questions...
UPS: APC1400RMXL, with an SNMP management card. IP address 192.168.3.2
Server 1: Dual P200, 2 EtherNet cards, Linux kernel Vers 2.2.14 with IPChains, IP addresses, 192.168.3.1, 199.85.233.247
Server 2: P75 Linux Kernel Version 2.2.10, IP address 192.168.3.60
Windoze1: P400 Win98, Whatsup Gold Vers 3.5, IP Address 192.168.3.53
WorkStation1: Same machine as above, but under Linux RedHat 6.0
WinDoze2: P266 Win98, Whatsup Gold Vers 3.5, IP address 192.168.3.50
WorkStation2: Same as above, but with Slackware 4.0
So... With that, the P400 when running under Windoze can both ping the UPS, and can get SNMP information using the WhatsUp gold SNMP probe utility. Now, when I boot up under Linux I can't ping the UPS, and I cant get any SNMP values, either by snmpget or snmpwalk. The same behaviour is exhibited on "WinDoze2", and none of the other Linux machines will ping, and none of them will do an snmpget. When I try to do an snmpwalk, I get the following (this will also provide the SNMP adapter version, etc):
nile:/usr/src/ucd-snmp-4.1.2# snmpwalk apc1400 public
system.sysDescr.0 = APC Embedded PowerNet SNMP Agent (FW v3.0.1 SW v2.2.4.a, HW B2, MOD: AP9605, Mfg: 03/19/1998, SN: WA9812744961)
Timeout: No Response from apc1400
nile:/usr/src/ucd-snmp-4.1.2#
Help! Please!
Thanks,
Richard
---
Message number 2:
| To: Derek Simkowiak <[EMAIL PROTECTED]> Subject: Re: Major frustrations... Cc: [EMAIL PROTECTED] |
| At 12:17 PM 5/8/2000 -0700, you wrote: |
| Please be warned first off that this is a LONG message!! |
-->So... With that, the P400 when running under Windoze can both ping the UPS,
-->and can get SNMP information using the WhatsUp gold SNMP probe utility.
-->Now, when I boot up under Linux I can't ping the UPS, [...]
Something about your message doesn't make sense to me.
Something about this whole *THING* doesn't make sense to me... Here's something that's really going to seem even weirder...
nile:/usr/src/ucd-snmp-4.1.2# telnet apc1400
Trying 192.168.3.2...
Connected to apc1400.connections.yk.ca.
Escape character is '^]'.
User Name :
telnet> close
Connection closed.
nile:/usr/src/ucd-snmp-4.1.2#
In this telnet session I've connected to the UPS from the same Linux box that can't SNMPWalk to it... This behaviour is consistent with all the other Linux boxes as well (They can Telnet, to the UPS, but can't ping it)... So.. Here's the results of my ping session with both the APC and another box on the same network...
nile:/usr/src/ucd-snmp-4.1.2# ping -c 10 192.168.3.60
PING 192.168.3.60 (192.168.3.60): 56 data bytes
64 bytes from 192.168.3.60: icmp_seq=0 ttl=255 time=1.3 ms
64 bytes from 192.168.3.60: icmp_seq=1 ttl=255 time=1.3 ms
64 bytes from 192.168.3.60: icmp_seq=2 ttl=255 time=1.0 ms
64 bytes from 192.168.3.60: icmp_seq=3 ttl=255 time=1.1 ms
64 bytes from 192.168.3.60: icmp_seq=4 ttl=255 time=1.2 ms
64 bytes from 192.168.3.60: icmp_seq=5 ttl=255 time=1.0 ms
64 bytes from 192.168.3.60: icmp_seq=6 ttl=255 time=0.9 ms
64 bytes from 192.168.3.60: icmp_seq=7 ttl=255 time=1.1 ms
64 bytes from 192.168.3.60: icmp_seq=8 ttl=255 time=0.9 ms
64 bytes from 192.168.3.60: icmp_seq=9 ttl=255 time=1.0 ms
--- 192.168.3.60 ping statistics ---
10 packets transmitted, 10 packets received, 0% packet loss
round-trip min/avg/max = 0.9/1.0/1.3 ms
nile:/usr/src/ucd-snmp-4.1.2# ping -c 10 192.168.3.2
PING 192.168.3.2 (192.168.3.2): 56 data bytes
--- 192.168.3.2 ping statistics ---
10 packets transmitted, 0 packets received, 100% packet loss
nile:/usr/src/ucd-snmp-4.1.2#
Here's the network settings dialogue from the APC:
> 1
------- TCP/IP ----------------------------------------------------------------
The Network Service has started with the following settings :
-----------------------------------------------------------------------
Adapter IP : 192.168.3.2
Subnet Mask : 255.255.255.0
Default Gateway : 192.168.3.1
MAC Address : 00 C0 B7 62 87 B2
1- Adapter IP : 192.168.3.2
2- Subnet Mask : 255.255.255.0
3- Default Gateway : 192.168.3.1
4- BOOTP : Disabled
5- Accept Changes :
?- Help
<ENTER> Redisplay Menu
<ESC> Return To Previous Menu
Here's a result of the PING test from the APC to the host that I can't ping or snmp FROM...
------- Ping Utility ----------------------------------------------------------
1- Address To Ping : 192.168.3.1
2- Ping : Pending
?- Help
<ENTER> Redisplay Menu
<ESC> Return To Previous Menu Without Accepting Changes
> 2
------- Ping Utility ----------------------------------------------------------
1- Address To Ping : 192.168.3.1
2- Ping : Ping Success
?- Help
<ENTER> Redisplay Menu
<ESC> Return To Previous Menu
...and a ping from the APC to another one of my hosts that has NO firewalling code of any stripe on it (since it exists entirely inside the network)...
> 1
Address To Ping : 192.168.3.60
------- Ping Utility ----------------------------------------------------------
1- Address To Ping : 192.168.3.60
2- Ping : Pending
?- Help
<ENTER> Redisplay Menu
<ESC> Return To Previous Menu Without Accepting Changes
> 2
------- Ping Utility ----------------------------------------------------------
1- Address To Ping : 192.168.3.60
2- Ping : Ping Success
?- Help
<ENTER> Redisplay Menu
<ESC> Return To Previous Menu
Now, I go to the host I just pinged (192.168.3.60), and try to ping the APC..
voyager:~# ping -c 10 apc1400
PING apc1400.connections.yk.ca (192.168.3.2): 56 data bytes
--- apc1400.connections.yk.ca ping statistics ---
10 packets transmitted, 0 packets received, 100% packet loss
voyager:~#
I have been doing networking for a LOT of years, I've seen some really bizarre things, and have managed to solve pretty much all of them, but to this day this one STILL has me stumped. My thought is it must be something in either the Linux Kernel, or somewhere else, and to that end I'll try the Linux-networking list as well, but this is doubly strange, because ping is of course an ICMP utility, and SNMP is (by default) a UDP utility. So, to that end, I put an extra SNMP entry in my /etc/services file to support TCP as well as UDP, and I get the following:
nile:/usr/src/ucd-snmp-4.1.2# snmpwalk -T TCP apc1400 public
snmpwalk: Invalid local port (Connection refused)
nile:/usr/src/ucd-snmp-4.1.2#
Any and ALL suggestions and help will be MUCH appreciated...
Thanks in advance,
Richard W.
Richard Whittaker on the road or @ work
ICQ UIN 1390720 - You can get ICQ at:
http://www.icq.com - Try
it today!
WWW Homepage:
http://ganges.connections.yk.ca/rwhittak
Work: 867-393-7756
