This is a rewrite of a userspace USB driver for TEMPer devices presenting
a USB ID like this: 0c45:7401 Microdia
My device came from M-Ware ID7747
and also reports itself as 'RDing TEMPerV1.2'.
Also provides a passpersist-module for NetSNMP (as found in the snmpd
packages of Debian and Ubuntu) to present the temperature of 1-3 USB devices
via SNMP.
Basically, libusb bindings for python (PyUSB) and snmp-passpersist from PyPI.
Under Debian/Ubuntu, treat yourself to some package goodness:
sudo apt-get install python-usb python-setuptools snmpd # The latter is only necessary for SNMP-usage.
sudo easy_install snmp-passpersist
After you run
sudo python setup.py install
you should end up with two scripts conveniently installed:
/usr/local/bin/temper-poll
/usr/local/bin/temper-snmp
If your system does not provide access as a normal user to the USB device, you need to rum them as root. See "USB device permissions" section for more on this.
temper-poll accepts -p option now, which adds the USB bus and port information each device is plugged on.
without -p option
$ temper-poll
Found 1 devices
Device #0: 22.5°C 72.5°F
with -p option
$ temper-poll -p
Found 1 devices
Device #0 (bus 1 - port 1.3): 22.4°C 72.3°F
Which tells you there is a a USB hub plugged (internally or externally) on the port 1 of the bus 1 of the host, and your TEMPer device is on the port 3 of that hub.
Regarding errors:
usb.core.USBError: [Errno 16] Resource busyUnknown error
By default, the Linux kernel claims (e.g. opens/uses) the TEMPer device as a keyboard (HID device). When that happens, this script is not able to set the configuration and communicate with it.
You will see one of those two errors when running sudo temper-poll. Your dmesg log will show something similar to this:
usb 1-1.3: usbfs: interface 0 claimed by usbhid while 'temper-poll' sets config #1
To prevent this, add this to the kernel command line:
usbhid.quirks=0x0c45:0x7401:0x4
On Raspbian, this will be /boot/cmdline.txt. Reboot after saving and retry. Hat tip to and more information from AndiDog here.
Using NetSNMP, you can use temper/snmp.py
as a pass_persist module.
You can choose one of two OIDs to be emulated: APC's typical
internal/battery temperature (.1.3.6.1.4.1.318.1.1.1.2.2.2.0) or
Cisco's typical temperature
OIDs
(.1.3.6.1.4.1.9.9.13.1.3.1.3.1 - 3.3).
Note that you should not activate both modes at the same time.
The reason for this limitation is that the script will keep running for each
pass_persist entry and they will interfere with each other when updating the
temperature.
This typically leads to syslog entries like this:
temper-python: Exception while updating data: could not release intf 1: Invalid argument
At least on Debian Wheezy, the default USB device node has permissions to only allow
access for root. In the same case, snmpd is running as the user snmpd. Bam. No access.
You might find a corresponding note in syslog.
To solve that, the file 99-tempsensor.rules is a udev rule that allows access to the
specific USB devices (with matching VID/PID) by anyone. Install like this:
sudo cp etc/99-tempsensor.rules /etc/udev/rules.d/
To check for success, find the bus and device IDs of the devices like this:
pi@raspi-temper1 ~ $ lsusb | grep "0c45:7401"
Bus 001 Device 004: ID 0c45:7401 Microdia
Bus 001 Device 005: ID 0c45:7401 Microdia
pi@raspi-temper1 ~ $ ls -l /dev/usb*
crw------- 1 root root 189, 0 Jan 1 1970 /dev/usbdev1.1
crw------- 1 root root 189, 1 Jan 1 1970 /dev/usbdev1.2
crw------- 1 root root 189, 2 Jan 1 1970 /dev/usbdev1.3
crw-rw-rwT 1 root root 189, 3 Jan 1 1970 /dev/usbdev1.4
crw-rw-rwT 1 root root 189, 4 Jan 1 1970 /dev/usbdev1.5
pi@raspi-temper1 ~ $
Note that /dev/usbdev1.4 and /dev/usbdev1.5 have permissions for read/write
for anyone, including snmp. This will work for the passpersist-module running
along with snmpd.
To emulate an APC Battery/Internal temperature value, add something like this to snmpd.conf. The highest of all measured temperatures in degrees celcius as an integer is reported.
pass_persist .1.3.6.1.4.1.318.1.1.1.2.2.2 /usr/local/bin/temper-snmp
Alternatively, emulate a Cisco device's temperature information with the following. The first three detected devices will be reported as ..13.1.3.1.3.1, ..3.2 and ..3.3 . The value is the temperature in degree celcius as an integer.
pass_persist .1.3.6.1.4.1.9.9.13.1.3 /usr/local/bin/temper-snmp
Add --testmode to the line (as an option to snmp.py to enable a mode where
APC reports 99°C and Cisco OIDs report 97, 98 and 99°C respectively. No actual devices
need to be installed but libusb and its Python bindings are still required.
The path /usr/local/bin/ is correct if the installation using python setup.py install
did install the scripts there. If you prefer not to install them, find and use the
temper/snmp.py file.
The error reporting of NetSNMP is underwhelming to say the least. Expect every error to fail silently without a chance to find the source.
snmp.py reports some simple information to syslog with an ident string
of temper-python and a facility of LOG_DAEMON. So this should give you the available debug information:
sudo tail -f /var/log/syslog | grep temper-python
Try stopping the snmpd daemon and starting it with logging to the console:
sudo service snmpd stop
sudo snmpd -f
It will not start the passpersist-process for snmp.py immediately
but on the first request for the activated OIDs. This also means that the
first snmpget you try may fail like this:
iso.3.6.1.4.1.9.9.13.1.3.1.3.2 = No Such Instance currently exists at this OID
To test the reporting, try this (twice if it first reports No Such Instance):
snmpget -c public -v 2c localhost .1.3.6.1.4.1.9.9.13.1.3.1.3.1 # Cisco #1
snmpget -c public -v 2c localhost .1.3.6.1.4.1.9.9.13.1.3.1.3.2 # Cisco #2
snmpget -c public -v 2c localhost .1.3.6.1.4.1.9.9.13.1.3.1.3.3 # Cisco #3
snmpget -c public -v 2c localhost .1.3.6.1.4.1.318.1.1.1.2.2.2.0 # APC
When NetSNMP starts the instance (upon first snmpget), you should see something like this in syslog:
Jan 6 16:01:51 raspi-temper1 temper-python: Found 2 thermometer devices.
Jan 6 16:01:51 raspi-temper1 temper-python: Initial temperature of device #0: 22.2 degree celsius
Jan 6 16:01:51 raspi-temper1 temper-python: Initial temperature of device #1: 10.9 degree celsius
If you don't even see this, maybe the script has a problem and quits with an exception.
Try running it manually and mimik a passpersist-request (-> means you should enter the rest of the line):
-> sudo temper/snmp.py
-> PING
<- PONG
-> get
-> .1.3.6.1.4.1.318.1.1.1.2.2.2.0
<- .1.3.6.1.4.1.318.1.1.1.2.2.2.0
<- INTEGER
<- 22.25
If you have a problem with the USB side and want to test SNMP, run the script with --testmode.
The devices I have seen do not have any way to identify them. The serial number is 0.
There is no way (and this driver does not make any attempt) to present a persistent
ordering among the USB devices. The effective order is the one that libusb presents.
That seems to be based on the enumeration order of the devices.
That in turn seems to be based primarily on the physical ordering in the root hub -> hub port hierarchy on bootup. But if you unplug and replug the device (or it gets detached due to a glitch and is redetected) then the order of the devices may be changed.
If that happens, your temperature readings will change and you cannot say which device belongs to what OID if you are using SNMP.
Long story short: Only use the device order if the USB bus is stable and you reboot after any plugging on the device. Even then, you are not safe. Sorry.
Since calibration parameters must be set per each device, we need some way to identify them physically. As mentioned above, the serial number for all TEMPer devices is zero, so there is no true way to tell which is which programatically. The USB device number does not work either since it changes every time you reboot the machine or plug/unplug the device. The way that possibly can work is identifying them by the combination of the bus number and the USB port (possibly a chain of ports, if you have hubs in between), which is what I am doing for now.
This information is basically the same with what you can get with lsusb -t and is based on the information in the sysfs directory /sys/bus/usb/devices (see below). So far I am assuming this scheme is persistent enough for regular use cases, but even the bus number may change in some cases like - for example - if your machine is a tablet like machine and you hotplug it to a keyboard dock with a USB root hub in it. In such case you will need to re-run lsusb and adjust the bus-port numbers in the configuration file accordingly. At the moment I have no clue about SNMP OID persistence.
You can have parameters in the configuration file /etc/temper.conf for each of your TEMPer device to calibrate its value with simple linear formula. If there is not this file on your machine it's fine, calibration is just skipped. The same if the program can't find a matching line with the actual device on the system.
Format of calibration lines in /etc/temper.conf is:
n-m(.m)* : scale = a, offset = b
where n is the USB bus number and m is (possibly a chain of) the USB port(s)
which your TEMPer device is plugged on. a and b are some floating values decided by experiment, we will come back to this later, first let me describe how n and m can be decided for your device.
You will need to use lsusb command in usbutils package to decide n and m. Use lsusb with and without -t option.
For example, assume the following outputs;
$ lsusb
Bus 002 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Bus 001 Device 016: ID 0c45:7401 Microdia
Bus 001 Device 015: ID 1a40:0101 TERMINUS TECHNOLOGY INC. USB-2.0 4-Port HUB
Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
$ lsusb -t
/: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=orion-ehci/1p, 480M
/: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=orion-ehci/1p, 480M
|__ Port 4: Dev 15, If 0, Class=hub, Driver=hub/4p, 12M
|__ Port 3: Dev 16, If 0, Class=HID, Driver=usbhid, 1.5M
|__ Port 3: Dev 16, If 1, Class=HID, Driver=usbhid, 1.5M
First output tells you your TEMPer device (0c45:7401 Microdia) is on the bus 1 and has (just currently, it may change time to time, even if you don't move it around) device ID = 16.
Now look at the second output. Looking at this tree, your TEMPer device (Dev 16) on the bus 01 is connected to your pc through two ports, port 4 and port 3. Don't worry about two devices having the same Dev ID = 16, they both belong to a single TEMPer device (it uses two USB interfaces by default, which is normal).
So in this example, n = 1 and m = 4.3; thus the config file should be like
1-4.3: scale = a, offset = b
with a and b replaced with the actual values which you will need to measure and
calculate for your own TEMPer device. These values are used in the formula
y = a * x + b
where
y: calibrated temperature (in Celsius),x: raw temperature read from your TEMPer device (in Celsius).
You will need to find appropriate values for a and b for your TEMPer device by doing some experiment and basic math. Either comparing it with another thermometer which you can rely on or measuring two temperatures which you already know ... like iced water and boiling water, but make sure in the latter case that you seal your TEMPer device firmly in a plastic bag or something, since it is NOT waterproof!
To find out bus and port numbers, you can also try running temper-poll with -p option, which will contain information in the form (bus 1 - port 4.3) in the above example. This might be actually easier than looking at the lsusb outputs, as long as it works.
The USB interaction pattern is extracted from here as seen on Google+.
This should work on both Python 2 and 3. It was tested with Python 2.7.3 and 3.2.3.
The snmp_passpersist mode is Python 2 only because the upstream package is not ready yet.
- Original rewrite by Philipp Adelt [email protected]
- Additional work by Brian Cline
- Calibration code by Joji Monma (@GM3D on Github)
- Munin plugin by Alexander Schier (@allo- on Github)
- PyPI package work and rewrite to
libusb1by James Stewart (@amorphic on Github) - Reduced kernel messages and support multiple sensors by Philip Jay (@ps-jay on Github)
- Python 3 compatibility and rewrite of cli.py to use argparse by Will Furnass (@willfurnass on Github)