raspberriesareyummy

joined 1 year ago
[–] raspberriesareyummy 1 points 16 hours ago (1 children)

For most of history you would be better off if you could kill the next village over.

That is an incredibly stupid take. For most of history, the planet was so vast that people had plenty of room to hunt / farm / whatever. And no, killing other humans is not in our DNA, the only people who feel like that are those with brain damage / development defects.

[–] raspberriesareyummy 2 points 19 hours ago (1 children)

again, I don't have a problem with copying code - but I as a developer know whether I took enough of someone else's algorithm so that I should mention the original authorship :) My only problem with circumventing licenses is when people put more restrictive licenses on plagiarized code.

And - I guess - in conclusion, if someone makes a license too free, so that putting a restrictive (commercial) license or patent on plagiarized / derived work, that is also something I don't want to see.

[–] raspberriesareyummy 0 points 19 hours ago (3 children)

As I am a big proponent of open source, there is nothing wrong even with copying code - the point is that you should not be allowed to claim something as your own idea and definitely not to claim copyright on code that was "inspired" by someone else's work. The easiest solution would be to forbid patents on software (and patents altogether) completely. The only purpose that FOSS licenses have is to prevent corporations from monetizing the work under the license.

[–] raspberriesareyummy 15 points 1 day ago (2 children)

“Why does no one say murder is bad unless China is murdering”

I can not fathom how you absolutely nailed the essence of my comment, yet misunderstood it (and - arguably - your own example) so fundamentally.

Let me try to help, once:

"Why do most people not complain about murder when Microsoft is doing it, but when China is doing it, the very justified outrage can be heard?"

[–] raspberriesareyummy 5 points 1 day ago (3 children)

I am also sick to the core about this aspect of humanity. I feel that we as a species are just about developed enough to understand how a better world would look like, and how people should act, what's "the right thing to do" - and very much not developed enough to overcome our egoism and narcissism to make it happen, so we do the wrong thing despite knowing better far too often.

[–] raspberriesareyummy 80 points 1 day ago (26 children)

With the obligatory "fuck everyone who disregards open source licenses", I am still slightly amused at this raising eyebrows while nearly no one is complaining about MS using github to train their copilot LLM, which will help circumvent licenses & copyrights by the bazillion.

[–] raspberriesareyummy 0 points 1 day ago

whatever you need to tell yourself to not have to self-reflect the bullshit you are spouting...

[–] raspberriesareyummy 0 points 1 day ago (2 children)

Words have meaning. And while it is absolutely and fundamentally wrong to impose religious bullshit on the education system (or anyone, really), no, you are moving the goal posts: The ten commandments have nothing to do with Nazis. You arguing against evangelical fundamentalists doing whatever has nothing to do whatsoever with my previous comment,

[–] raspberriesareyummy 19 points 1 day ago (5 children)

Also, the winners will interprete who gets to be the axis and who gets to be the allies in history books...

[–] raspberriesareyummy 3 points 2 days ago (1 children)

...said the little troll after referring the ssuper sspecial sseckrit translation that only it knows... Yeah, suck it, loser.

[–] raspberriesareyummy 3 points 2 days ago (3 children)

Have you considered that maybe I know something you don’t?

Considered briefly, but as you decided not to share the alleged "apartheid" commandments, I dismissed that consideration as unlikely and yourself as a troll.

[–] raspberriesareyummy 1 points 2 days ago

oh - my apologies, I forgot that on-board graphics have a dedicated chipset. Also, no idea whether on-board sound would have used CPU power back in the late days of soundcards, as the comment I responded to was claiming... might have been a sound chip for that, too..

 

I am on a binge, listening to tons of Nigerian, Ugandan and Kenian music, and I am absolutely loving it. Why is this not playing on our (Western European) radio stations regularly?

Instead I have to regularly change stations because someone is trying to torture my ears with Ed Sheeran or similar BS. :(

 

Hadn't seen this before - I wonder if that was once on the potato outside and then grew closed?

107
submitted 4 months ago* (last edited 4 months ago) by raspberriesareyummy to c/[email protected]
 

Ich bin nicht mit Allem einverstanden, was mein 18-jähriges Ich verbrochen hat, aber für diese Gewissenserklärung werde ich mich niemals schämen.

 

Hey everyone! I wanted to post my guide / experiences from the past years here - this is not strictly Zabbix, but SNMP (which Zabbix could monitor), and in C++, but I think given that lemmy isn't yet huge, I'd rather have the post in a space with more topics, than an isolated 1-user-community. See below.

End 2022, I had finally gotten around to creating a standalone SNMP agent with a lot of demo-functionality. I will post the link / instructions for that in a separate comment below.

#Getting started with net-snmp in C++

I am trying to work myself into the use of net-snmp (on a Linux system, but ideally in a way compatible to mingw/gcc on Windows). Unfortunately, the net-snmp documentation as per http://www.net-snmp.org/ seems to be very lacking. There are coding tutorials, which at some point refer you to an empty section in the wiki that you "should read before continuing" (This one here: Writing_a_MIB_Module). So I am down to a lot of web (re)searching and trial & error.

The aim of this post is to end up with a step-by-step approach that accomplishes the following:

  • Installation & configuration of all required software packages on a debian system (for starters)
  • Clarification of the relevant snmp terminology used in this step-by-step guide
  • Making the example executables (or modified versions thereof) from the net-snmp.org coding tutorials work to
    • request a specific value by existing snmp object identifiers (OIDs) with a targeted instruction
    • request a list of OIDs listed in an array, from multiple hosts
    • create & use a custom management information base (MIB)
    • create a custom snmp agent that can serve requests for OIDs defined in that custom MIB
    • create a simple monitoring application that can cyclically request certain OIDs from the custom agent

The final demo implementation will be a "hello world" of sorts for SNMP, wherein the custom snmp agent reports it's application "health" (e.g. process id, process name, a timestamp and a cyclic counter / heartbeat) via the custom MIB, and the monitoring application monitors the health of that custom agent on the console (as opposed to having to set up s.t. more complicated like zabbix).

Please bear with me, as I am stuck on the custom agent for now. I have to solve this problem, however, and I will update this post with my solution. Help is very much appreciated!

So - without further ado, here is my current status:

Package installation:

# force installation of snmpd with default /etc/snmp/snmpd.conf
sudo apt install --reinstall -o Dpkg::Options::="--force-confask,confnew,confmiss" snmpd

# force installation of default /etc/snmp/snmp.conf
sudo apt install --reinstall -o Dpkg::Options::="--force-confask,confnew,confmiss" libsnmp-base

# install snmp and snmp-mibs-downloader
sudo aptitude install snmp snmp-mibs-downloader
# Note: former net-snmp-utils files are now all included in package snmp

Package configuration:

# comment out the "mibs :" line in snmp.conf (TBC: forgot why I did this :( )
sudo sed -i 's/^mibs :$/#mibs :/' /etc/snmp/snmp.conf

# nothing else for the moment

Activate configuration changes (if any)

sudo /etc/init.d/snmpd restart

#Test package configuration:

snmpwalk -v 2c -c public localhost
# a bunch of output about localhost will follow

Explanation: snmpwalk checks the destination (localhost) for all available OIDs and requests them once.

Note: Option values can be separated by space or not at all, so -v 1 and -v1 is equivalent.

Option -v is the snmp protocol version, -v1 or -v2c work. Disclaimer: -v3 requires authentication logic that I have not yet figured out. Option -c sets the "community" string - which I have not quite figured out yet, but "public" is good for accessing OIDs that do not require a password.

Speculation: I believe the community string is simply correlated to how certain OIDs are registered in the snmp master agent (daemon) and you can only access them through the configured community.

Manual request of an OID on the command line:

snmptest -v 2c -c public localhost

This will open a request prompt, on which you can enter the OID / variable identifier of anything you previously saw on the output of snmpwalk. Multiple variables can be requested at once. E.g.:

Variable: SNMPv2-MIB::sysDescr.0
Variable: SNMPv2-MIB::sysORID.1
Variable: 

In order to submit (send) the request, you have to enter an empty line. When you are done, exit snmptest with CTRL-C. Now that we are done with the installation & configuration, it is time to get the demo applications running.

Getting ready for the net-snmp coding tutorial:

I have downloaded the following files from the coding tutorial subsections here:

asyncapp.c
example-daemon.c # called example-demon.c there, I renamed it
Makefile
NET-SNMP-TUTORIAL-MIB.txt
nstAgentModuleObject.c
nstAgentModuleObject.h
nstAgentPluginObject.c
nstAgentPluginObject.h
nstAgentSubagentObject.c
nstAgentSubagentObject.h
snmpdemoapp
snmpdemoapp.c

These files were not created by me and I take no credit for them. I have, however, modified these files to fit my personal preference of C code formatting, and to add comments and better understand what is going on - and also to fix some incompatibilities with my system and change them to use SNMP_VERSION_2c so I could get a working software before I have to study the authentication logic of version 3.

I have uploaded my modified version of the tutorial files here (link valid only until 2020-05-20, I'll upload it elsewhere if someone knows a good free filehoster): 2020-05-06-net-snmp-coding-tutorial.tgz

Please review the code before compiling stuff some random stranger gave you ;)

You need to copy the tutorial MIB file to a place where your snmpd (the master agent) will find it:

sudo cp NET-SNMP-TUTORIAL-MIB.txt /usr/share/snmp/mibs/

Testing whether the MIB file is found by the snmp applications:

# snmptranslate -m +NET-SNMP-TUTORIAL-MIB -IR -On nstAgentModuleObject
# (the above has the same effect as the below)
snmptranslate -On NET-SNMP-TUTORIAL-MIB::nstAgentModuleObject

This should output

.1.3.6.1.4.1.8072.2.4.1.1.1

Compiling and executing the tutorial files:

Disclaimer: As of today (2020-05-05), I have only managed to successfully use the snmpdemoapp and the asyncapp - the other apps are still untested / non-functional in this version.

When you execute "make" in the folder with your source files & the Makefile, you will get two warnings

example-daemon.c:50:3: warning: implicit declaration of function ‘init_vacm_vars’; did you mean ‘init_vacm’? [-Wimplicit-function-declaration]
   init_vacm_vars();
   ^~~~~~~~~~~~~~
   init_vacm
example-daemon.c:51:3: warning: implicit declaration of function ‘init_usmUser’; did you mean ‘init_usm’? [-Wimplicit-function-declaration]
   init_usmUser();
   ^~~~~~~~~~~~
   init_usm

I have not yet figured out if this code is intentionally referring to unknown functions, and whether or not this is functional or a failure. Let's ignore it for now.

After building the executables, you can test:

./snmpdemoapp 
SNMPv2-MIB::sysORDescr.3 = STRING: The SNMP Management Architecture MIB.
value #1 is a string: The SNMP Management Architecture MIB.

and

./asyncapp 
---------- synchronous -----------
23:25:47.095474 localhost: SNMPv2-MIB::sysDescr.0 = STRING: Linux YOURHOSTNAME <YOUR OS VERSION AND ARCHITECTURE>
23:25:47.095637 localhost: SNMPv2-MIB::sysORDescr.1 = STRING: The MIB for Message Processing and Dispatching.
23:25:47.095758 localhost: SNMPv2-MIB::sysORDescr.2 = STRING: The management information definitions for the SNMP User-based Security Model.
23:25:47.095961 localhost: SNMPv2-MIB::sysDescr.0 = STRING: Linux YOURHOSTNAME <YOUR OS VERSION AND ARCHITECTURE>
23:25:47.096076 localhost: SNMPv2-MIB::sysORDescr.1 = STRING: The MIB for Message Processing and Dispatching.
23:25:47.096199 localhost: SNMPv2-MIB::sysORDescr.2 = STRING: The management information definitions for the SNMP User-based Security Model.
---------- asynchronous -----------
23:25:47.096538 localhost: SNMPv2-MIB::sysDescr.0 = STRING: Linux YOURHOSTNAME <YOUR OS VERSION AND ARCHITECTURE>
23:25:47.096657 localhost: SNMPv2-MIB::sysDescr.0 = STRING: Linux YOURHOSTNAME <YOUR OS VERSION AND ARCHITECTURE>
23:25:47.096769 localhost: SNMPv2-MIB::sysORDescr.1 = STRING: The MIB for Message Processing and Dispatching.
23:25:47.096877 localhost: SNMPv2-MIB::sysORDescr.1 = STRING: The MIB for Message Processing and Dispatching.
23:25:47.096986 localhost: SNMPv2-MIB::sysORDescr.2 = STRING: The management information definitions for the SNMP User-based Security Model.
23:25:47.097036 localhost: SNMPv2-MIB::sysORDescr.2 = STRING: The management information definitions for the SNMP User-based Security Model.

Now you can modify the snmpdemoapp.c and the asyncapp.c to include / request different hosts / OIDs. Please note that I have requested the same OIDs from localhost twice in asyncapp.c to allow this tutorial executable to function without any further host ready on the network and to demonstrate the sync vs. async behavior.

Edit 2020-05-06: Progress - I got a subagent registered (for now, only as root user :( ) and running as expected:

Configure a community "tutorial" that we can use to access our tutorial MIB

# configuring a community with access rights to custom MIB(s)
# Source: https://unix.stackexchange.com/questions/463969/snmp-no-such-object-available-on-this-agent-at-this-oid

# it appears that the common path is .1.3.6.1 and then custom MIBs have the .4 branch (TBC)

# add a custom community permitting read access to our tutorial MIB
echo "" > new-communities-for-snmpd.conf
echo "# custom communities" >> new-communities-for-snmpd.conf
echo "rocommunity tutorial  127.0.0.1 .1.3.6.1.4 # TBC: branch for custom mibs" >> new-communities-for-snmpd.conf
cat new-communities-for-snmpd.conf | sudo tee -a /etc/snmp/snmpd.conf
rm new-communities-for-snmpd.conf

Restart the snmpd daemon to load the new config:

sudo service snmpd restart

Build tutorial executables:

make

Attempt to retrieve an OID handled by the example daemon

snmpget -v 2c -c tutorial localhost NET-SNMP-TUTORIAL-MIB::nstAgentSubagentObject.0

Since the daemon is not yet running, this will fail: "NET-SNMP-TUTORIAL-MIB::nstAgentSubagentObject.0 = No Such Object available on this agent at this OID"

Start the tutorial subagent:

sudo ./example-daemon -f

Open a second console and attempt to retrieve the OID as above:

snmpget -v 2c -c tutorial localhost NET-SNMP-TUTORIAL-MIB::nstAgentSubagentObject.0

This time, the request should succeed: "NET-SNMP-TUTORIAL-MIB::nstAgentSubagentObject.0 = INTEGER: 2"

Check the first terminal (where the example-daemon is running), it should say

hello world!

This message was triggered by the handler function handler_nstAgentSubagentObject installed in nstAgentSubagentObject.c

/* Function: handler_nstAgentSubagentObject
 * Purpose: callback handler for the request to an OID of this subagent
 * 
 */
int handler_nstAgentSubagentObject(
	netsnmp_mib_handler *mibHandler,
	netsnmp_handler_registration *handlerRegistration,
	netsnmp_agent_request_info *agentRequestInfo,
	netsnmp_request_info *requestInfo
)
{
	printf( "hello world!\n" );

	return 0;
}

The handler function is registered as a callback handler in the call to netsnmp_register_long_instance.

From here on, this is all work in progress. Next steps:

  • create a handler routine that will work with snmpwalk & allow to see all OIDs identified for a certain community
  • create a custom MIB with sample application telemetry
  • create a handler routine that will update the OID values as part of the snmp agent (e.g. by interacting with a separate application thread that keeps updated telemetry in a struct)
  • create a monitoring application that periodically checks the content of the custom MIB and displays it on the console
  • somewhere along the way: figure out how to run all this as non-root user!!

I stand by my original comment: The state of documentation of this de-facto standard in worldwide networking equipment has me baffled. It is embarrassingly shitty.

view more: next ›