NeDi Community

NeDi Software Specific => Discovery => Topic started by: mduling on February 20, 2009, 09:07:00 PM

Title: Feasible for NeDi to track HP printer information?
Post by: mduling on February 20, 2009, 09:07:00 PM
Would it be feasible for NeDi to have a feature that tracked HP network printers, so that it could have a few key graphs of printer info?  Perhaps a Node => Printers
or something such that pages printed, toner level, etc. would be listed, as are the 3 Cisco graphs (mem, temp, cpu) are for devices.

Not sure if this is feasible, or if so if it would dilute the focus of NeDi, but I'm curious.  Thanks!
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rickli on February 21, 2009, 12:53:33 PM
It wold definitely not be NeDi's focus to monitor printers, but on the other hand....amazing potential! As I work for (kind of) a printer company now it even makes sense  ::)

You'll only need to find the proper OIDs, add Hewlett to ouidev and make sure descfilter plays along as well and off you go. You can even monitor the printers then, have them send syslog messages to NeDi (if they can do that) or show printers on maps. The more I think about it the more I like the idea, as this would broaden NeDi's clientbase to printadmins as well! Hmm, a NeDi-PRT edition falls into mind...

My 4000N doesn't seem to be that good with SNMP but at least I could specify a syslog host:

Code: [Select]
  ===JetDirect Telnet Configuration===
Firmware Rev. : G.07.17
MAC Address : 00:60:b0:f3:f7:bb
Config By : USER SPECIFIED

IP Address : 10.10.10.20
Subnet Mask : 255.255.255.0
Default Gateway : 10.10.10.1
Syslog Server : 10.10.10.11
Idle Timeout : 90 Seconds
Set Cmnty Name : Not Specified
Host Name : HERMES

Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on February 21, 2009, 07:57:45 PM
I'll look into the OIDs.  I assume only you would be able to do the coding for querying the OIDs and graphing them, correct?  Major job ... er I mean donation?  :)

Would work like this - poll for say 6 OIDs (if that many are useful) and graph say 3 (if only three are graph-worthy)?  I'm having a hard time picturing how this would look exactly.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rufer on February 23, 2009, 08:02:23 AM
You could already do something with existing nedi code.

Just add the 3 (4) interesting OID under the DEF file graph OID (instead of CPU/mem/temp). Check the MIBs of the printer for other interesting stuff like serial number etc and add that to the DEF file as well.
I do that for other devices as well.

Greetings
Rufer
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on February 23, 2009, 07:42:54 PM
Ah I see, I'll look into that.  Thanks.  I had forgotten about the sysoids.  NeDi is awesome.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on February 23, 2009, 08:50:36 PM
Okay wait a second.  The sysobj definitions "RRD graphing" section seems not general enough for non-switch/router devices.  For example, the items defined there are not combined into one graph and hard-coded to be type COUNTER.  It would be nice if that were generalized.  Here's an idea.

Have the "RRD graphing section" expanded to contain a discrete graphs (default)/ combined graphs decision and a COUNTER / GAUGE (default) decision.  That would allow for a lot more flexibility for non-cisco devices that is needed.  For printers, say toner levels for green/blue/red/whatever in the same graph as an example for combined graphs (and type GAUGE), and say page count as an example of type COUNTER (combined or discrete).

What do you think?
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rufer on February 24, 2009, 09:10:04 AM
This could be indeed useful for some special cases.
On the other hand, I understand that Remo tries to keep it as "simple and stupid" as possible.

Greetings
Rufer
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rickli on February 24, 2009, 06:20:40 PM
Rufer's statement is my credo, as there's enough complex stuff out there  ;)

Anyhow, not all rrds are counters. Just look at the system rrds...but maybe I didn't quite get what you ment. But why focus on graphs? This is NeDi not Cacti  ;) I had a go at "expanding" Device-Status for print supplies (see attached). It would be easy to generate a report of top empty toners etc. While I want to keep NeDi's main focus on network devices, this would be an interesting benefit for more general IT admins (pure networkers don't have to discover them, hence shouldn't be bothered). Printers then are treated as any other SNMP-enabled network infrastructure device.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rauchi on February 25, 2009, 05:14:39 PM
looks quite interesting.

Marco
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rickli on February 25, 2009, 06:24:33 PM
BTW I've been thinking about custom graphing as well (for years actually) and while hacking some .defs to provide associations on Wlan APs etc. is quite dirty, I think I have a solution though.

Coming from Cisco I've provided a dedicated value for IO mem. My question to all of you is, how important is this to you? Don't worry I'm not going to take it away, but make it customizable to graph more valuable information, if applicable...
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on February 27, 2009, 07:07:44 AM
I didn't think about the device status enhancing for hp printers like that.  What other details might there be if you add this printer functionality?  Total pages printed?

As far as the graphing, I can't say for sure.  You're right, it isn't Cacti.  Since I know next to nothing about the coding involved or rrdtool (I just use rrd frontends) I don't know the complexity or possible complications.  In my imagination -less you know the easier it seems :) - I supposed that maybe the additions I hinted at was just a matter of passing a few more parameters.  If it increased the development burden for you to support or evolve nedi then maybe it isn't worth it.  If it were not a burden from that perspective then it would be cool in theory to have the ability to stack graphs and/or do COUNTER graphs in a sysobj definition.  I don't know how much value there would be in fact; perhaps others can comment.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rufer on February 27, 2009, 01:52:18 PM
Personally I think it could be useful to have either
The possibility to add fully customizable graphs (choose title and type counter/gauge/...)
or
be able to change title for the existing graphs.

As example I use the existing graphs for other things like:
CPU as humidity graph (temp/humidity sensors)
temperature as "number of users connected" graph
Mem as jitter graph (some VoIP adapters)
Mem as voltage graphs (UPS)
CPU load as % output load (UPS)
CPU load as Receiver signal strengt (dBm) for some wireless links

This works really well for me (I know what it means) but it doesn't mean anything to others. For me, 3 graphs (4 values) are sufficent for most of my devices.
Of course it is a bit "quick&dirty", as example Nedi believes my Bluesocket overheats big time because there are 400 users connected and that value is graphed in the temperature graph ;D

Greetings
Rufer
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on February 28, 2009, 12:35:59 AM
Rufer, can you explain what you mean by 3 graphs but 4 values?  Also I have not tried to add a non-standard graph to the RRD graph section, and I'm having trouble doing it.

I assumed from the blank MemIO entry in the sysobj definitions that a fourth graph is possible.  True or not?  Because I added another OID and renamed that to "Conn" and no additional graph appeared (I deleted the current rrds and the device in the DB to start from scratch.  So then I repeated the process (deleted my asa) leaving the MemIO in as the label.  Still no connections graph.

So is a 4th OID possible in nedi and if not what is the 4th MemIO label for?  Thanks!

 
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rufer on March 02, 2009, 08:21:06 AM
You can use MemIO and MemCPU as two values in one graph.
I think the values must be of the same type as a memory graph. So no interface counters there, only absolute values.

It works really well. If you like, I can post some examples with DEF entry.

Greetings
Rufer
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 02, 2009, 06:21:45 PM
Yes, that would be great if you could give an example.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 02, 2009, 11:11:25 PM
Well I replaced the temp OID with the OID for connections (1.3.6.1.4.1.9.9.147.1.2.2.2.1.5.40.6) in my ASA sysobj but I see a number way lower than what it should be and not what the snmpget value is for that OID.  Isn't that all I'm supposed to have to do?  I want to do what you did in your "temperature as "number of users connected" graph" example.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rufer on March 03, 2009, 09:08:52 AM
OK, here are some examples

bluegw.conn.png
Bluesocket. Number of connections in temperature graph.
The device supports maximum 400 connections at a time, so yesterday we were in trouble.
DEF entry:
Temp    1.3.6.1.4.1.9967.1.13.1.1.0

unidreg.voltage.png
APC AP7721 rack automatic transfer switch (2 power sources, 1 common power output). Voltage graph in CPU graphs.
As you can see one voltage is stable 230V, thus from USV, the other fluctuates a bit, thus non-USV.
DEF entry:
MemCPU  1.3.6.1.4.1.318.1.1.8.5.3.3.1.3.1.1.1
MemIO   1.3.6.1.4.1.318.1.1.8.5.3.3.1.3.2.1.1


unidsensor.humidity.png
AKCP SP2 temperature/humidity sensor. Ambient humidity% in CPU graph.
DEF entry:
CPUutl  1.3.6.1.4.1.3854.1.2.2.1.17.1.3.0

One important thing: You always have to keep in mind how RRD treats these values. Read the RRD FAQ. It always depends on RRDstep and in which frequency RRD runs! I have a RRDstep of 3600 and nedi (and thus RRD) runs every 1h, so the values are right.
If you have a big rrdstep and run RRD manually every 5min for testing, the values will probably be wrong.

Greetings
Rufer
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 03, 2009, 07:57:23 PM
Cool thanks.  Well I'm getting no joy with using temp for my ASA (firewall) connections.  Here is my graphing section:

# RRD Graphing
CPUutl  1.3.6.1.4.1.9.9.109.1.1.1.1.5.1
Temp    1.3.6.1.4.1.9.9.147.1.2.2.2.1.5.40.6
MemCPU  1.3.6.1.4.1.9.9.48.1.1.1.6.1
MemIO

The connections run between 15,000 to 40,000.  I see messages like this in the messages list "Temperature of 15665 degrees Celsius detected!" but my graph is blank.  I would think auto-scaling would mean large numbers would not be a problem for rrdtool, but maybe nedi.pl chokes on a high value for temperature.

Title: Re: Feasible for NeDi to track HP printer information?
Post by: rickli on March 03, 2009, 09:42:10 PM
I'm halfway through migrating the IO Memory graph into it's own one and make it custom "lablable"...is that even a word? Anyway I'm aware it's only one value for now, but we'll see further once I get some feedback...

As for the Temp GAUGE, you're just a bit too eager :) Thinking a fith of the sun's surface temperature would be enough I limited it to +/- 1000 degrees (I know, the - is even more difficult): "DS:temp:GAUGE:$ds:-1000:1000"

Try the existing IO Memory instead, as it's from 0 to unlimited "DS:memio:GAUGE:$ds:0:U"
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 04, 2009, 10:54:05 PM
Using MemIO did it, thanks!  And thanks for your work on the change for added flexibility.
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 05, 2009, 01:22:52 AM
On my ASA tracking connections is so nice in nedi along with CPU because you can see the effect on the CPU of the connection load.  I think this would be a natural addition to the ASA sysobj definition.

I think I already contributed an ASA 5520 (1.3.6.1.4.1.9.1.670.def) sysobj file.  So how would that file need to be for the update you have planned to track connections (1.3.6.1.4.1.9.9.147.1.2.2.2.1.5.40.6)?  It would be cool if an updated sysobj file could be included in the next update.

Title: Re: Feasible for NeDi to track HP printer information?
Post by: rickli on March 05, 2009, 07:07:23 PM
I'm afraid I don't have that .def (669 and 674 are there). Mail it to me and I'll include it...
Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 05, 2009, 08:38:42 PM
Did you get 875 & 877?  http://forum.nedi.ch/index.php?topic=333.0

Well here is what I have for 670.  The icon may not be correct but the interfaces and modules works well.


# Definition for 1.3.6.1.4.1.9.1.670 created by Defgen 1.4 on 2.Dec 15:45:09 (admin)

# General
SNMPv   2HC
Type    ASA5520
OS      IOS-fw
Icon    vpbn
Bridge
Dispro
Serial  1.3.6.1.2.1.47.1.1.1.1.11.1
Bimage  1.3.6.1.2.1.47.1.1.1.1.10.1

# Vlan Specific
VLnams
VTPdom
VTPmod

# Interfaces
IFalia
IFalix
IFvlan
IFvlix
IFdupl
IFduix
Halfdp
Fulldp

# Modules
Modesc  1.3.6.1.2.1.47.1.1.1.1.2
Moclas  1.3.6.1.2.1.47.1.1.1.1.5
Movalu  9
Moslot  1.3.6.1.2.1.47.1.1.1.1.7
Modhw   1.3.6.1.2.1.47.1.1.1.1.8
Modsw   1.3.6.1.2.1.47.1.1.1.1.9
Modfw   1.3.6.1.2.1.47.1.1.1.1.10
Modser  1.3.6.1.2.1.47.1.1.1.1.11
Momodl  1.3.6.1.2.1.47.1.1.1.1.13

# RRD Graphing
CPUutl  1.3.6.1.4.1.9.9.109.1.1.1.1.5.1
Temp    1.3.6.1.4.1.9.9.392.1.3.1.0
#Workaround to get Current Connections in the graph labelled MemIO
#MemCPU 1.3.6.1.4.1.9.9.48.1.1.1.6.1
MemIO   1.3.6.1.4.1.9.9.147.1.2.2.2.1.5.40.6

Title: Re: Feasible for NeDi to track HP printer information?
Post by: mduling on March 05, 2009, 08:53:01 PM
BTW, how do you tell which def file a device is using?  I think I had to copy this

1.3.6.1.4.1.3854.1.2.2.1.1.def

to this

1.3.6.1.4.1.3854.1.def

to get my ACKP sensorProbe2 to work.  I would like to confirm which def file it is using, but since I left the original file in place, I want to be sure since I am going from memory.  I also don't understand why the one def file for AKCP has "Type  akcp-HHMSS".
Title: Re: Feasible for NeDi to track HP printer information?
Post by: rufer on March 06, 2009, 09:54:45 AM
The DEF file is chosen upon the SNMP sysOID of a device.
AKCP decided to change those between software versions. That's why there are multiple DEF files for them.

Greetings
Rufer