SCO OSR3 SNMP mibs ?
Stan Brown
OID's the SCO OSR3 SNMP agent is capable of supplying data on?
I have a mixed network that I am setting up mrtg to monitor, and one of
the boxes is ann SCO OSR3 box.
Thanks,
Jeff Liebermann
> Can anyone tell me where to find some information as to what MIBs &
> OID's the SCO OSR3 SNMP agent is capable of supplying data on?
Just the usual SMI and Hostmib stuff. See:
http://www.cruzio.com/~jeffl/sco/snmp_install.txt
Note that all of the problems mentioned were finally fixed in 3.2v5.0.6.
> I have a mixed network that I am setting up mrtg to monitor, and one of
> the boxes is ann SCO OSR3 box.
OSR3 would be 3.2v4.something. You may have problems using mosy and
post_mosy trying compile additional mibs. I'm still running SNMP on my
3.2v4.2 box in my palatial office without problems.
--
Jeff Liebermann je...@comix.santa-cruz.ca.us
150 Felker St #D Santa Cruz CA 95060
831-421-6491 pager 831-429-1240 fax
http://www.cruzio.com/~jeffl/sco/ SCO stuff
Stan Brown
>On 2 Nov 2000 13:36:10 -0500, st...@panix.com (Stan Brown) wrote:
>> Can anyone tell me where to find some information as to what MIBs &
>> OID's the SCO OSR3 SNMP agent is capable of supplying data on?
>Just the usual SMI and Hostmib stuff. See:
> http://www.cruzio.com/~jeffl/sco/snmp_install.txt
>Note that all of the problems mentioned were finally fixed in 3.2v5.0.6.
Ah, thanks, that page is a big help.
>> I have a mixed network that I am setting up mrtg to monitor, and one of
>> the boxes is ann SCO OSR3 box.
>OSR3 would be 3.2v4.something. You may have problems using mosy and
>post_mosy trying compile additional mibs. I'm still running SNMP on my
>3.2v4.2 box in my palatial office without problems.
3.2v5.0.5 if it matters.
Thanks for the help, Sounds like I can get all the stuff I need, once I
get a good snmd.conf file set up. Interesting most othher vendors will
repy to some things withou even a snmpd.conf file.
Thanks for the help.
Jeff Liebermann
>>OSR3 would be 3.2v4.something. You may have problems using mosy and
>>post_mosy trying compile additional mibs. I'm still running SNMP on my
>>3.2v4.2 box in my palatial office without problems.
>
> 3.2v5.0.5 if it matters.
OSR3 is Open Desktop 3.2v4.x and such.
OSR5 is OpenServer 5 as in 3.2v5.0.x.
For the exact version and OS patches installed, run:
customquery listpatches | head -1
> Thanks for the help, Sounds like I can get all the stuff I need, once I
> get a good snmd.conf file set up. Interesting most othher vendors will
> repy to some things withou even a snmpd.conf file.
There's a reason. SNMP2, 2.1 and now 3 primarily deal with security
issues. These are usually buried in multiple convoluted snmpd.conf
files. Under some (not all) Linux mutations, UCD-SNMP arrives where
even localhost cannot read the local machine without first tweaking
the security stuff. I guess this is an improvement. Anyway, in
3.2v5.0.6, mkdev snmp was cleaned up with reasonably secure defaults,
while 3.2v5.0.5 and before are wide open on arrival.
Incidentally, if there's some stuff you wanna monitor, (sar
statistics, idle precentage, memory usage, diskspace), I have some
shell scripts for the purpose which will work for just about anything
worth graphing. I keep promising Tobias to document it properly.
Sigh, yet another project.
Also, be advised that the 3.2v5.0.6 version has:
brand -u
which will return the number of licenses consumed. Very handy when an
aspiring user decides to see how many telnet sessions he can open and
runs the system out of licenses.
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831.336.2558 voice
# 831.426.1240 fax http://www.cruzio.com/~jeffl
# 831.421.6491 digital_pager je...@comix.santa-cruz.ca.us
Scott Neugroschl
Stan Brown
>On 2 Nov 2000 15:21:24 -0500, st...@panix.com (Stan Brown) wrote:
>Incidentally, if there's some stuff you wanna monitor, (sar
>statistics, idle precentage, memory usage, diskspace), I have some
>shell scripts for the purpose which will work for just about anything
>worth graphing. I keep promising Tobias to document it properly.
>Sigh, yet another project.
Y would love to see this, if you don't mind.
>Also, be advised that the 3.2v5.0.6 version has:
> brand -u
>which will return the number of licenses consumed. Very handy when an
>aspiring user decides to see how many telnet sessions he can open and
>runs the system out of licenses.
Now that's an interesting sidelight. Let me tell you the bacjground
there, if you don't mind.
The machine in question was provided by a vendor. When i was given
responsibility for it, one of the first request was to get it on the
network. Thinking it would be like any other *NIX machine, I walked up
to it, only to discover that SCO had unbundled netwking.
So, I called SCO, and tried to purchase the networking add on. They
would not sell it to me, saying I had to buy yhe whole OS again. I
grited my teeth, and started getting inof on how to do this. During
this converstaion (with a salse type), I was asked how many user
licenses I would need.
I said, like all my other *NIX bixes 2. The I though a minute and
decided to make certain that SCO had not done something strnage jer. To
clarify on all other *NIX's the network logins (any qty) only count as
one user. So 2 user lincenses will allow you to be loged on to the
console, and allow an esentially unlimited number of telnet etc.
netwrok session.
I was told that SCO did not play the game this way. Thinking that this
was pretty strange, since the history of this goes all the way back to
the AT&T license, I asked to talk to someone with a technical
background to verify this. Well I wound up talking to a technican, who
swore on a stack of manuals that it was like the salse gut sad.
So, I forked over the money for an extended user license pak (5 as I
recall). Well when I did the install, I did not bother to install this
lincense pack, and as I suspected I have never been refuseda telnet
session yet.
So, if you don't mind, what is the real story here?
Jeff Liebermann
Pleeeze don't mix two topics in the same thread.
> Y would love to see this, if you don't mind.
Very preliminary. [Insert disclaimer for lack of organization and
explanation]. Note that the general form can be used to extract anthing
that belches integers. The catch is that the integer function in MRTG
2.9.x seems to have been dropped in favor of having rateup.exe do the
trick so that RRDTOOL will work with fractions. Therefore, one could
probably use non-integers.
================
#!/bin/sh
# @(#) idle.sh Belch current idle percentage for mrtg graphing.
# Remember to have two output numbers on different lines.
#
# Typical output of sar -u 1
# 21:25:20 %usr %sys %wio %idle
# 21:25:21 0 4 5 92
#
# Really disgusting way to get rid of extra leading spaces
# by feeding it to a shell variable. Retch.
bletch=`sar -u 1 | tail -1`
echo $bletch | cut -d" " -f5
echo $bletch | cut -d" " -f5
echo "/n/n"
================
# Local sar idle time for mrtg.cfg.
Target[idle]: `idle.sh`
Title[idle]: Sar Idle Percentage
PageTop[idle]: <H1>Sar Idle Percentage</H1>
Options[idle]: growright, gauge
MaxBytes[idle]: 100
#WithPeak[idle]: dwmy
Unscaled[idle]: dwmy
YLegend[idle]: Idle
ShortLegend[idle]: Idle
================
This works nicely on the local machine and will graph the
idle percentage. No big deal. However there are other ways
to get the same results from a remote machine.
Modify the bletch= line to read:
bletch=`rcmd $1 "sar -u 1 | tail -1"`
which will run the sar -u 1 command on a remote system
passed as arguement $1. The line in the mrtg.cfg file
would read:
Target[192.168.111.1]:`idle.sh 192.168.111.1`
to get idle statistics from 192.168.111.1. This requires
some security games with hosts.equiv.
================
I used the same method to extract diskspace info:
#!/bin/sh
# by Jeff Liebermann 04/15/98
#
# Get various diskspace numbers for MRTG.
#
# This is the format belched by df -v -i
# Mount Dir Filesystem blocks used free %used iused ifree
%iused
# / /dev/root 1050000 972132 77868 93% 59872 71384 46%
# /stand /dev/boot 30000 16414 13586 55% 14 3746 1%
# /u /dev/u 600000 252560 347440 43% 4259 70741 6%
# /usr/spool /dev/news 184492 5830 178662 4% 8 23056 1%
#
# Really disgusting way to get rid of extra leading spaces
# by feeding it to a shell variable. Ugly at best.
#
# usage: whatever machine Filesystem
# i.e. whatever comix /dev/root
#
retch=`rcmd $1 "df -v -i $2" | tail -1 | tr -d %` # just one
Filesystem
set $retch # break apart into fields using IFS seperators
echo $6 # Percentage diskspace used
echo $9 # Percentage inodes used
echo "/n/n"
#
================
> The machine in question was provided by a vendor. When i was given
> responsibility for it, one of the first request was to get it on the
> network. Thinking it would be like any other *NIX machine, I walked up
> to it, only to discover that SCO had unbundled netwking.
Turned around, one might suggest that there are a substantial number of
machines which do not require networking. As SCO has to play licence
fees for each TCP/IP runtime, the savings to the customer is substantial.
> So, I called SCO, and tried to purchase the networking add on. They
> would not sell it to me, saying I had to buy yhe whole OS again.
That's odd. You apparently have 3.2v5.0.5 "host" which arrives as a 5
user, no networking package. No need to buy a new runtime.
http://www.UniXpress.com/sco/unix/openserver/5/SCOOpenServerUpgrades.html
Host to 3.2v5.0.6 enterprise upgrade. XA261-UX50-5.0 $799 list price.
> grited my teeth, and started getting inof on how to do this. During
> this converstaion (with a salse type), I was asked how many user
> licenses I would need.
That's odd. The tcp/ip package picks up the license count from the OS.
This sales droid was clueless. Approximately how many years ago was this
as SCO Service Sales has improved substantially in the last year and
would normally not make such a mistake.
> I said, like all my other *NIX bixes 2. The I though a minute and
> decided to make certain that SCO had not done something strnage jer. To
> clarify on all other *NIX's the network logins (any qty) only count as
> one user. So 2 user lincenses will allow you to be loged on to the
> console, and allow an esentially unlimited number of telnet etc.
> netwrok session.
Well, here we run into a problem. I almost lost a customer because I had
to shove a $500 license count upgrade down their throat to avoid database
crashes that occurred when the license count was exeeded. See:
http://www.sco.com/cgi-bin/ssl_reference?107560
and note that it applies to only 3.2v5.0.0. The plan was to count
"keyboards". I'm not sure it worked as documented. There were also some
changes in 3.2v5.0.6 to the way licenses were counted. I don't recall
the exact details. It's in the release notes somewhere.
What's this about a 2 user? What *EXACTLY* did you have before upgrading
to "enterprise"? If it's the free 2 user demo system, you pay full tilt
for the upgrade.
> I was told that SCO did not play the game this way. Thinking that this
> was pretty strange, since the history of this goes all the way back to
> the AT&T license, I asked to talk to someone with a technical
> background to verify this. Well I wound up talking to a technican, who
> swore on a stack of manuals that it was like the salse gut sad.
He/She/It was wrong. When was this?
> So, I forked over the money for an extended user license pak (5 as I
> recall). Well when I did the install, I did not bother to install this
> lincense pack, and as I suspected I have never been refuseda telnet
> session yet.
Well, it would be interesting to know how many users the system is
licensed for. Run:
uname -a
> So, if you don't mind, what is the real story here?
Dunno. I usually stay out of this kind of debate. If you hadn't tacked
your SCOSS and license manager question onto an SNMP thread, I wouldn't
even have bothered to read it.
--
Jeff Liebermann je...@comix.santa-cruz.ca.us
150 Felker St #D Santa Cruz CA 95060
Stan Brown
>On 4 Nov 2000 10:15:35 -0500, st...@panix.com (Stan Brown) wrote:
>Pleeeze don't mix two topics in the same thread.
>> Y would love to see this, if you don't mind.
>Very preliminary. [Insert disclaimer for lack of organization and
>explanation]. Note that the general form can be used to extract anthing
>that belches integers. The catch is that the integer function in MRTG
>2.9.x seems to have been dropped in favor of having rateup.exe do the
>trick so that RRDTOOL will work with fractions. Therefore, one could
>probably use non-integers.
Thansk, I am studyng it.
>Turned around, one might suggest that there are a substantial number of
>machines which do not require networking. As SCO has to play licence
>fees for each TCP/IP runtime, the savings to the customer is substantial.
Well the lasy UIX box I bought without networking was my Fortune 32:16
:-)
Short test to see how long you have been in this biz.
>> So, I called SCO, and tried to purchase the networking add on. They
>> would not sell it to me, saying I had to buy yhe whole OS again.
>That's odd. You apparently have 3.2v5.0.5 "host" which arrives as a 5
>user, no networking package. No need to buy a new runtime.
That's the replacemant OS. I don't have the old one to get a bersion
number from.
>http://www.UniXpress.com/sco/unix/openserver/5/SCOOpenServerUpgrades.html
>Host to 3.2v5.0.6 enterprise upgrade. XA261-UX50-5.0 $799 list price.
>
>> grited my teeth, and started getting inof on how to do this. During
>> this converstaion (with a salse type), I was asked how many user
>> licenses I would need.
>That's odd. The tcp/ip package picks up the license count from the OS.
>This sales droid was clueless. Approximately how many years ago was this
>as SCO Service Sales has improved substantially in the last year and
>would normally not make such a mistake.
This was about 10 months ago.
>> I said, like all my other *NIX bixes 2. The I though a minute and
>> decided to make certain that SCO had not done something strnage jer. To
>> clarify on all other *NIX's the network logins (any qty) only count as
>> one user. So 2 user lincenses will allow you to be loged on to the
>> console, and allow an esentially unlimited number of telnet etc.
>> netwrok session.
>Well, here we run into a problem. I almost lost a customer because I had
>to shove a $500 license count upgrade down their throat to avoid database
>crashes that occurred when the license count was exeeded. See:
> http://www.sco.com/cgi-bin/ssl_reference?107560
>and note that it applies to only 3.2v5.0.0. The plan was to count
>"keyboards". I'm not sure it worked as documented. There were also some
>changes in 3.2v5.0.6 to the way licenses were counted. I don't recall
>the exact details. It's in the release notes somewhere.
>What's this about a 2 user? What *EXACTLY* did you have before upgrading
>to "enterprise"? If it's the free 2 user demo system, you pay full tilt
>for the upgrade.
No, it was a purchased one. I really don;t remeber the version number.
I can check at work Mondau, as I still have the old machine siting in
my office. I got the ompession that it was only 1 rev back, tho.
>> I was told that SCO did not play the game this way. Thinking that this
>> was pretty strange, since the history of this goes all the way back to
>> the AT&T license, I asked to talk to someone with a technical
>> background to verify this. Well I wound up talking to a technican, who
>> swore on a stack of manuals that it was like the salse gut sad.
>He/She/It was wrong. When was this?
Again, about 10 months agao.
>> So, I forked over the money for an extended user license pak (5 as I
>> recall). Well when I did the install, I did not bother to install this
>> lincense pack, and as I suspected I have never been refuseda telnet
>> session yet.
>Well, it would be interesting to know how many users the system is
>licensed for. Run:
> uname -a
>> So, if you don't mind, what is the real story here?
>Dunno. I usually stay out of this kind of debate. If you hadn't tacked
>your SCOSS and license manager question onto an SNMP thread, I wouldn't
>even have bothered to read it.
Sorry, somehow you raised the issue in your original reply. Thats what
got me going on it.
Apologies, and thnaks for the SNMP help.
Stan Brown
>On 2 Nov 2000 13:36:10 -0500, st...@panix.com (Stan Brown) wrote:
>> Can anyone tell me where to find some information as to what MIBs &
>> OID's the SCO OSR3 SNMP agent is capable of supplying data on?
>Just the usual SMI and Hostmib stuff. See:
> http://www.cruzio.com/~jeffl/sco/snmp_install.txt
>Note that all of the problems mentioned were finally fixed in 3.2v5.0.6.
Finally got a chance to set this p today.
Thanks for all teh help on it. I have a couple of things to mention
hee. Your writeup seems to imply that starting up the host mib subagent
will satrt snmpd itself. At least in my case it did not. So I had to
start it before runing the rc script for the host subagent MIB.
Second, tha data available from this MIB seeems to be lacking some
pretty basic things. I don't see average CPU load, or disk I/O rates,
or used memory, or used swap space for instance.
Are ther other MIBs that I can tunr on, to get some more data?
Agains, thanks for your help with this.
Jeff Liebermann
> Thanks for all teh help on it. I have a couple of things to mention
> hee. Your writeup seems to imply that starting up the host mib subagent
> will satrt snmpd itself. At least in my case it did not. So I had to
> start it before runing the rc script for the host subagent MIB.
Huh? I'm highly partial to having my instructions followed in the order
presented. First, you get snmpd running using mkdev snmp. That should
setup starting snmpd in /etc/rc.d/something. Once that's working, you
install the VTCL hostmib stuff with mkdev hostmib, which is a seperate
daemon. No clue where you got the idea that the order can be reversed.
> Second, tha data available from this MIB seeems to be lacking some
> pretty basic things. I don't see average CPU load, or disk I/O rates,
> or used memory, or used swap space for instance.
Well, if you want 1400 lines of output (I just counted) that my Caldera
Linux SNMP belches, methinks you may have the opposite problem. Yeah,
hostmib is insipid, mosy and post_mosy are old, it's only SNMP V1, and
there's lots of pieces missing.
> Are ther other MIBs that I can tunr on, to get some more data?
No. That's why I wrote and posted the shell scripts that belch sar, df,
and ps data in a form suitable for MRTG managment and graphing.
Stan Brown
>On 6 Nov 2000 13:35:44 -0500, st...@panix.com (Stan Brown) wrote:
>> Thanks for all teh help on it. I have a couple of things to mention
>> hee. Your writeup seems to imply that starting up the host mib subagent
>> will satrt snmpd itself. At least in my case it did not. So I had to
>> start it before runing the rc script for the host subagent MIB.
>Huh? I'm highly partial to having my instructions followed in the order
>presented. First, you get snmpd running using mkdev snmp. That should
>setup starting snmpd in /etc/rc.d/something. Once that's working, you
>install the VTCL hostmib stuff with mkdev hostmib, which is a seperate
>daemon. No clue where you got the idea that the order can be reversed.
I am pretty certain that doing a stop on the host mid daemon stoped the
main snmpd daemin. If you are certain I am wrong, then don't worry
about it. I was trying to follow your directions in "cookbook" fashin,
tho. Not a big deal, if you are ceratin. It's just that your directions
weer _so_ good, I thought that if indeed I had found a small
discrepency you might waht to correct it.
>> Second, tha data available from this MIB seeems to be lacking some
>> pretty basic things. I don't see average CPU load, or disk I/O rates,
>> or used memory, or used swap space for instance.
>Well, if you want 1400 lines of output (I just counted) that my Caldera
>Linux SNMP belches, methinks you may have the opposite problem. Yeah,
>hostmib is insipid, mosy and post_mosy are old, it's only SNMP V1, and
>there's lots of pieces missing.
>> Are ther other MIBs that I can tunr on, to get some more data?
>No. That's why I wrote and posted the shell scripts that belch sar, df,
>and ps data in a form suitable for MRTG managment and graphing.
Mmm, that aren't all that useful, if you are collecting data over
thenetwork from the SCO box, as oposed to runing mrtg on the SCO box,
right, or am I mising something?
Jeff Liebermann
> Mmm, that aren't all that useful, if you are collecting data over
> thenetwork from the SCO box, as oposed to runing mrtg on the SCO box,
> right, or am I mising something?
Sigh. Allow me to introduce you to the wonders of "rcmd". This lets you
execute a remote program (in this case sar or df), and return the results
to a data conglomerator (MRTG). To setup rcmd, you will need to tweak
/etc/hosts.equiv and possibly /.rhosts. These have the potential to be
major security holes so do it carefully. If you can run the remote
program as something other than root (such as UID=mrtg), then do so.
Instead of running the shell script locally to get sar statistics as in
my example as:
bletch=`sar -u 1 | tail -1`
you use rcmd to run it on the remote system as in:
bletch=`rcmd $1 "sar -u 1 | tail -1"`
(The $1 passes the IP address of the target machine to the shell script).
Play with rcmd, learn by destroying, and give it a try. It works on
about a dozen machines I have it installed.
It's also possible to add a service to inetd.conf that runs a script and
use telnet to return the results. I've had some problems with this as
telnet refused to exit if it encounters an error. Instructions, if you
really wanna try it.
Tony Lawrence
> It's also possible to add a service to inetd.conf that runs a script and
> use telnet to return the results. I've had some problems with this as
> telnet refused to exit if it encounters an error. Instructions, if you
> really wanna try it.
You can get better control of telnet by wrapping it in expect,
but really what you want is a nice little client-server app that
can send what you want and act upon it at the other end.
Fortunately, you already have a couple of these handy little
things: one is called "mail" and the other is remote printing:
see http://www.aplawrence.com/SCOFAQ/scotec1.html#clientserver
--
Tony Lawrence (to...@aplawrence.com)
SCO/Linux articles, help, book reviews, tests,
job listings and more : http://www.pcunix.com