I am looking for a solution to remotely power cycle embedded devices that are connected via a USB hub to a server. I have a software solution (usbreset.c) that was posted in several responses in stack overflow - rebooting the usb port. This works if the device is still alive, can be detected by the server, but, it just fails to communicate with the server. I have had to use it a few times. however, sometimes the device could hang or freeze on a gui page and there is no serial communication. This requires physically power cycling the device using it's on-off switch. Someone suggested raspberry-pi solution. But as per that solution, I would need one raspberry-pi per device. It's as if using a raspberry-pi to serve as a relay. This is expensive and cannot scale. Could relays work in this situation i.e. the power pins of each device connect to a relay. So, one relay per device. There are remote IP power cycle solutions. so via a web interface it is possible to check which device is offline and power cycle just that one. Such solutions are common in IT. However, is such a solution viable to remote power cycle embedded devices connected at the other end?
Related
I am currently building an app that controls a single actuator and a LED strip. The controller for the actuator and LED strip is based on an ESP32.
On the ESP32 I have used the <ESPAsyncWebServer.h>, <ESPAsyncWiFiManager.h> to set up and connect the ESP to the local wireless network. In the app I would like to send some commands to the ESP like http://<ESP32_IP>/led_set?level=<level> and
http://<ESP32_IP>/act_open and so on, this shouldn't be an issue.
Normally I would set up mDNS on the ESP and in the Flutter app, however several of the ESP controllers are already shipped to the customers. So I am looking for another way to identify the ESP.
At the moment I have used the ping_discover_network library and the ESP shows up fine on port 80. I am however not able to separate the ESP from the other devices that shows up on port 80.
Do any of you have any suggestions?
Thank you in advance
You don't have many good options. If you only want to identify a few devices which have slipped through your fingers without mDNS, you can use their MAC address as a filter. If you happen to know their individual MAC addresses, then that should be quite painless. Otherwise you'd have to look for devices with MAC OUI (first 3 bytes, usually) ranges allocated to Espressif. Espressif has several OUI ranges, but if your devices were from a single batch of ESP32 module of the same type, they'll be using the same OUI so it should be quite easy to figure out.
E.g. a batch of modules I have all start with 40:F5:20:...
Of course, if a third party installs their (unrelated) ESP32 module into the same network, you'll have a chance of identifying those as well, but such is life.
This is obviously a stop-gap measure. Make sure further devices go out of door with some identification (mDNS or otherwise).
I am trying to reverse engineer the API of an IoT device using mitmproxy. My setup is an iphone, computer running on MacOS 10.14, and an IoT device (watering pump) that can only access the wireless network after being plugged into the computer via USB connection, i.e. a not genuinely WiFi device. My phone is configured to point to the computer, which is running mitmproxy on a standard configuration.
When I send instructions from the app controlling the device on my phone to the device itself, presumably these instructions are sent to the computer, to the device cloud server, and then to the device. With these assumptions, one would think that they would see the flow of POST requests in mitmproxy before observing the results of those instructions. That is, if you send an instruction to turn on the pump, you'd think you'd see POST request containing that instruction show up in the mitmproxy flows before you see the pump turn on.
However that is not the case here. What happens is that, when I send instructions from the app, I observe the expected behavior from the IoT device, and then the flow of requests appear on the mitmproxy console seemingly at random. There seems to be no determinate relationship between the instructions I send and the requests that appear; they show up 5 seconds later, 5 minutes later, or 30 minutes later. Is this an intentional security feature? To somehow jam MITMproxies so that hackers cannot easily isolate the knowledge of which packet is performing which instruction? Or is it just something that I am doing wrong? Does anyone have any ideas as to what could be happening hear and potential solutions for making the flow of requests appear in real time? Ty
I am trying to detect if my iPhone is in the same network as my Raspberry Pi. I would like to execute a script when I am at home and my iPhone's presence is registered in my LAN.
It seems that when the phone is in standby not even the iphone-sync port (6207/tcp) is found. "/usr/bin/nmap -n -sT -p62078 [my phone's local IP]" shows no host. I wonder what else I could scan for. Obviously the phone is online and ready to accept facetime calls (data via 3G is deactivated). Could I accomplish something with avahi which I am using on my Raspberry Pi, or are there other ways.
I've just spent a week beating on this problem so I can refrain from sending SMS home alarms to my wife when she's at work.
Pinging won't work because the iPhone won't respond to ICMP when asleep. Reading the ARP cache won't work because a sleeping iPhone will come and go (check it every 30 seconds for a few minutes).
The only way I have found to 'reliably' determine when my two iPhones are on my local (home) network is to use the PCAP dotnet library to look for any packets originating from either of the phones' MAC addresses. For example, if you run Wireshark with the capture filter
ether src <iphone-mac-address>
you will see a surprising amount of network discovery/announcement traffic from the phone. It still has quiescent states, but so far the longest interval I have seen between captured packets is around 10 minutes. You would have to wait until you have not heard from the phone for some interval (I use 15 minutes) before declaring it not-home.
With this technique you will find a phone quickly when it rejoins the home network, assuming your phone is configured for DHCP. I also use port mirroring on my main Ethernet switch to include traffic from my wireless access points.
I don't have a Raspberry Pi solution for this, because my linux expertise is very limited, but someone else may be able to help you along those lines. I have a Windows Service using the PCAP library and so far it works reliably, with the limitation of waiting 15 minutes before deciding an iPhone has left the network.
* update 2-3-2018 *
I have this detection algorithm down to about 5 minutes, using a combination of ping/arp messages directed to each phone, about once per minute. Seems to work great.
You can find a list of devices on your network by investigating your arp cache.
arp -a
Simply write a bash script to run arp -a at a regular interval, and search for the mac address of your phone.
You could go even further with this and perform different actions depending on what brand of device is connected.
The first 3 hexadecimal digits of a mac address are the vendor id.
Take the following mac address:
00:19:E3:AB:CD:EF
00:19:E3: is one of the registered mac address for apple devices.
By comparing the devices on your network with this list, you could detect when for example a '3com' device, or a 'dell' device attaches to your network.
http://www.coffer.com/mac_find/?string=apple
You can do "arp-scan -l -r10" for that (tested this myself), but the problem is if mobile data enabled the iphone will go and suspend wifi if screen is locked to safe battery. so you need to disable mobile data .. then arp-scan will work.
There are free projects/examples of wol (wake on LAN) for iOS?
I've found this one [openwol][1] but it's old and has no signs of recents updates.
Also I've been digging on it and it's not working as expected, the main purpose of it is wake a computer but it's not working. I've spent some time on it and still no work.
Maybe there are other examples or someone else may join on me and solve/update this code to work?
Wake-on-LAN generally doesn't work for machines on wireless networks, as the wireless hardware is typically powered down when the machine is off or asleep -- it's usually only supported for wired Ethernet. As such, wake-on-LAN is unlikely to give acceptable results on an iOS device, as it'll only work for some specific network configurations (i.e, networks where the wireless segment is bridged to a wired segment that the target system is connected to).
Am I missing something glaringly obvious or is there no way to debug an iOS app which uses an external accessory that's connected via the 30-pin dock without using a bucket load of logs etc. I want to be able to use things such as breakpoints and Instruments.
Is there a way to remote debug perhaps, over Wi-Fi, Bluetooth?
Note: Yes, I asked this very recently and I deleted it because I thought I found the answer.. but the answer was only Instruments has support over Wi-Fi.. not Xcode debugging. So the question still remains...
And so...: Given that I've had no real luck finding the answer, and no one has given me an answer as yet - I take it that it is a big fat NO. :(
Makes me wonder are we just expected to magically guess where bugs occur, or log the crap out of everything while wearing out our dock connectors by continuously moving it back and forth between the device and accessory?
Time to file a bug report I guess.
At CES today, I talked to a developer from Wahoo Fitness that makes an ANT+ accessory for iPhone. They had this same problem, but found a solution.
They found a pass-through dock extender that has a mini-USB port. They used the mini-USB port for debugging while the accessory was connected.
The product they were using is http://www.cablejive.com/products/dockStubz.html
This blog talks about remote debugging iOS with a dock accessory attached
You could connect the external accessory to another iOS device (not the one tethered to the Mac running the Xcode debugger). Then tunnel all your EA framework messages from the accessory connected device to the device running the app being debugged over a pair of wifi sockets. Look at the code for tunneling accelerometer messages from a device to the iOS Simulator (a common trick for debugging game code on the Simulator) for one example of how this could be done.
After further researching, and having seen that people had to do sending strings over Wi-Fi to get around this, I'm concluding the answer is no.
I have filed a bug request for this.
In the mean time, it seems like the Wi-Fi logging, and on-device text logging will be the way to go for now.
Here's my understanding for why just the USB protocol works for some external accessories and doesn't for other external accessories. Looks like a fundamental problem, without an arbitrator, two masters can't talk to a single slave over USB, a serial Master/Slave protocol. So XCode is one master, the iPhone is the slave device. If the external accessory is a master too, one won't be able to connect the iPhone (Xcode slave) to the second master (the external accessory).
Probably the Wahoo Key for iPhone" is a slave device and that's why the dockStubz solution works for such an external accessory.
I have tested the dockStubz. It doesn't work for my external accessory. As suspected, the USB protocol can't be used to have two Master devices controlling a single slave device. Trying to hook up a Mac (Master) (via the mini USB ) & an external accessory (Master) (via the 30 pin connector) to the iPhone 4 (Slave) causes the iPhone to go in loop of connect & re-connect.
Following looks promising too, though expensive: digi.com/support/kbase/kbaseresultdetl.jsp?id=485.
Has any one tried to use USB to Ethernet connectors and use a router to route requests from two masters (XCode & External Accessory) to the slave (iPhone)? I am off to Best Buy to purchase USB to Ethernet cables and hook all three on to my IP router. Will report if it works.
This is what will be needed :
http://www.bestbuy.com/site/IOGEAR+-+USB+Ethernet+Extender/9614781.p?id=1218131339965&skuId=9614781&st=USB%20to%20Ethernet&cp=1&lp=1
http://www.frys.com/product/6103339
So connect XCode mac using the male end into the USB slot of your computer. . Use a ethernet cable to connect this to a router.
Connect the iPhone to the female part of the IO gear connector. Connect it to the router via ethernet cable.
Connect the external accessory with the male connector (Sabrent USB to Fast Ethernet Network Adapter.) Connect it to router.
I am still researching if this will work. Just ordered the parts. Will get it by Friday & will report back then.
Update:
The IOGear male end draws too much current when connected to router. Also, the female end can't charge iPhone when connected to the router even when 5V USB current supplied.
So tried to directly connect the iPhone to the USB slot of the router (used for printers). It does charge the iPhone. Also used USB to Fast Ethernet Network Adapter (BestBuy had one to connect Wii via USB) to connect the Mac to the router. It did connect to internet but couldn't find the iPhone. In the router client list I don't see any login entires for the iPhone. So this experiment was a failure unless someone have other pointers.