KVM with only Pi Zero W

[jtoland]

Hi,

I don't know if you are intending to update this but I am currently looking at getting this running on a single pi zero w. You can now get a CSI to HDMI adaptor so that everything can just run off 1 pi.

I was wondering if you could help me get this working if you have some spare time.

Kind Regards

[SterlingButters]

@jtoland myself and some collaborators are currently working on this. Hope to have some results soon. Will also be adding mouse and mass storage support. However you may find that a pi zero may not have sufficient computational capacity to process audio/video. Since the Pi4 supports OTG, this is a better option.

[jtoland]

Hello,

Thank you for confirming that you are doing this with the pi 4.

i have tested video on the pi zero and this seems to work fine with keyboard but having mouse working as well i can can see where the bottleneck could be.

Do you have any kind of time frame on the pi 4 version as i would love to test this even if it just has basic features such as keyboard only to start off with.

Kind Regards,

[calebcoverdale]

I'm excited to see some updates on the Pi 4 version.

My boss wants me to order a Raritan Dominion, but I would love to make this little project.

Do you have a recommendation on the hardware that would be needed for the Pi 4 version so I can get those ordered in ahead of time?

[SterlingButters]

The goal of our project is require only the Pi4 and a video capture device. There are some oddities with the CSI video capture on the Pi4 so we have chosen to go with a USB capture for now (most should work).

You can track my progress here but once it is working I may close source: https://github.com/SterlingButters/ip-kvm-interface/tree/dev?files=1

Currently I am able to capture keyboard and mouse input. All that's left is parsing the captured data to send to the target PC (assuming no issues with video capture). From there it'll likely just be some fine tuning that's needed.

[jtoland]

The issue I see with the pi 4 is the power consumption.

This would be ok for computer with USB 3 or type c but this would not work well with servers as the output is mainly USB 2.

That's why I have chosen to go with the zero.

I have had a look at the link you provided and will have to test it at some point.

Thanks for sharing.

[mdevaev]

@jtoland @SterlingButters @calebcoverdale
I have a very similar project in almost production-ready stage (but without documentation lol). My implementation is able to capture video using HDMI and CSI-2, switching video modes, hardware encoding on GPU and is capable of outputting more than 30 fps. Right now I can use any version of pi (started from 2) and one arduino with atmega32u4. I have a very advanced backend in which you can implement plugins to support OTG. For easy development I created a small toolkit that is able to assemble and install the required configuration of the operating system with just one command.

Perhaps we could combine our efforts?

https://github.com/pikvm

More photos: https://yadi.sk/d/X2DZP5Hr34uwyz

[SterlingButters]

@mdevaev Nice project! Definitely not opposed to combining efforts however I'm fairly confident that my code will not mesh well with yours. The intent of my project structure focuses on everything being very lightweight - using reliable existing libraries and features of node/js. Additionally, even though the project will support the Pi4 (maybe Pi0 because also supports OTG), it will not require any additional hardware besides maybe a USB capture card (working on B101 HDMI CSI2 bridge). Because the project is so lightweight, installation and documentation will be exceptionally easy. Unfortunately I've just been without wifi for several days so I haven't been able to put forth the progress I'd like to. Maybe you could detail some of the advantages of your project - I notice in your picture something about mass storage? Obviously no one wants to abandon their own project but if you are familiar with node/js I'd be happy to bring you aboard.

[mdevaev]

@SterlingButters
In fact, my project does not require specific hardware. Everything that you saw in the photos is optional, and can either be turned off or reconfigured in such a way as not to use additional equipment except for a video capture device. I did a lot of work so that any possible hardware configuration could be described with a simple file, and the user could use standardized APIs, interfaces and much more, regardless of how the KVM works inside :)

A few features that I already have:

• Microservice architecture - the system consists of a some separated parts that perform a strictly defined task.
• Control backend with clear API that can be used for scripts and alternative UI (for example, you can make a desktop application);
• Own MJPG streamer written on C than can change the resolution in real time, report statistics about the video and much more (see details in README).
• Ability to use any video capture device.
• Extensible authorization methods - you can configure multiple KVMs so that they use a common password database.
• IPMI BMC support (you can use ipmitool for power management in your existing network infrastructure).
• Mass-storage device based on flash drive (but! I can implement OTG for backend). Now I use a regular USB drive whose contacts switch between PI and computer using relay.
• One arduino for keyboard and mouse. The firmware implements a protocol with a check for transmission errors - it can never just hang. Can be replaced by OTG.
• A ready-to-use operating system that can be assembled using make build and installed to flash using make install.
• All components in packages have their own repository. In addition, I developed a special tool that allows you to build the operating system as if it were a regular docker container. It is very convenient and makes debugging easier.

Languages: C, Python, pure JS for Web UI.

I plan to implement support for the VNC and some other things - this fits easily into my architecture due to its simple extensibility.

[jtoland]

@mdevaev
Send me a link to your one-line setup and i will have a play and let you know.
Also what is the git hub page

[mdevaev]

@jtoland in a couple of days I'll write a small instruction for you

[Spamm00r]

Hi guys,

I'm very interested in your projects. And I'm also looking for a low cost generic IP KVM solution.

But I have issues with Raspberry and I prefer to go another route.

The problem with raspberry is that it is very selective for the hardware. I have 2 capture devices that I can not use because there are no drivers available for raspberry. Also using raspberry requires linux skills. And I as many other users are only able to copy and paste commands others have put in their tutorials. If something isn't working, I'm stuck. Like the missing drivers for my capture cards. Someone with skill may be able to create the drivers, but unfortunately I'm a Linux dummy.

If I now start buying all the right hardware for raspberry that is plug and play compatible, i'll end up with a higher bill than just buying a server Mainboard with vPro or IPMI. ALso right now with all the hardware required for the solutions posted, the Bill is somewhere above 100$.

Also the dozens of cables required to make the connection is problematic and error prone.

I wanted to cut costs and use an old vPro system instead of a raspberry. Old system having vPro capability can be bought for less than 30$, which is cheaper than a rapsberry. The advantage is that it is more powerful than a raspberry, and is managable out of band and you can run any OS you want on it, and if needed boot it bare metal. Thats perfect.

Right now my HTPC Raspberry system keeps freezing every other day even with a watchdog running and needs a manual hardreset by pulling the plug. VPro system however are rock solid and reliable. But of course only very few systems have vpro capabilities.

Of course the ultimate Aim is to control another system out of band that has no out of band capability. In my case I wanna be able to control AMD systems or non vPro Intel systems. For example non Q Chipset systems and be completely flexible about the system I use and not stuck to Q-Chip-based Boards and CPU's.

I have already several cheap vpro capable system, that I can VNC into on Bios level.

I only need Video Input and remote keyboard functionality to control any other pc.

I can handle Video Input in VLC on windows, but I have trouble controlling other system keyboard.

I have looked at several options I think that the Pi0 method you are employing is the best solution.

I have read that using an Arduino is cheaper, but I have trouble finding proper documentation about it.

Can you guys advice me on porting your Keyboard control method to windows?

I'm sure its pretty easy to connect the Pi0 to a PC via USB or Serial Port and use it as a Keystroke Passthrough solution.

What hardware do i need to connect to the Pi0?

I'd prefer to use a Pi0 WH and SSH into it wirelessly and get Keystroke pass through capability via ssh window without any wired connection between my vpro system and the Pi0. Is that possible?

It would be awesome if someone could describe the steps for that.

[mdevaev]

@jtoland https://github.com/pikvm/docs
If you have any problems or questions, you can contact us using Discord: https://discord.gg/bpmXfz5

[mdevaev]

@Spamm00r You must understand that any IP-KVM, as far as it would not be advanced, can never be more functional than the IPMI built into the motherboard since it has more access to the low-level functions of the hardware. If you need the cheapest IP-KVM, you can buy a compatible video capture device (from any of our projects) and use it.

I'd prefer to use a Pi0 WH and SSH into it wirelessly and get Keystroke pass through capability via ssh window without any wired connection between my vpro system and the Pi0. Is that possible?

I understand correctly that you want to have just a wireless IP-KVM, consisting of a minimum number of parts?

[Spamm00r]

@mdevaev Yes. All I need is a reliable method to pass through keystrokes with minimal hardware requirements and preferably wireless to the Target machine at BIOS level.

I was thinking that it should be possible to attach a Pi0 to the USB port of a Target and then SSH into that wireless Pi0 to issue Keystrokes to the target.

There are "Bad USB" devices like a: Wireless HID CJMCU-3212 that may achieve this. They are wireless and emulate keyboards. But I can not find a proper documentation on them.

I would however prefer a Pi0 as it is more versatile and this project here has already achieved this with a Pi0. And both devices cost about 15$.

I understand that i will be missing advanced features such as USB/PCI redirection etc. (I can achieve this with a bootable USB stick attached to the Target at all times) But that*s OK for me. All I want is a simple and cheap possibility to issue keyboard commands at BIOS level to the target.

For resetting the target, i can use a App controlled power socket to hardreset the target if the need arises. So clean and simple wireless solution. Just plug the power cord in the smart power socket and you can start and reset the Target pc.

I know how I would obtain the Video from the target with a capture device. Thats the easy part.
I think i have figured out all other parts.

However the Keystrokes over SSH to the Target is the difficult part. I need help with that.

The project you are working on looks awesome. But it looks overkill for my purposes. Also Your hardware list is quite long. may I ask what the hardware costs would be for the setup you are using?

[mdevaev]

@Spamm00r

May I ask what the hardware costs would be for the setup you are using?

~$100 or less. We plan to make a minimum implementation on Zero Pi. It will not be too fast, but probably in a couple of months we will be able to provide something.