I recently got a new 3D printer (Ender 3 Pro), and thought I would put up some of the small things I have recently printed. In trying to print things from Thingiverse, I couldn’t login even after making an account. I would get a spinning “Logging in” and it would never end. After looking at the network log, I saw it trying to reach out to https://accounts.thingiverse.com/unverified?username=danberk If you run into this issue, go to that URL with your username and it will send you an email to verify your account. Then the site will allow you to login.
I have been using Ruckus ICX 7150-12P switches at home recently, I wanted to have it more out of the way; so I designed and printed a mount that would mount the switch to the side. It came out well and looks good! I also printed a network cable comb to hold all the cables nicely together.
Having a small home lab I wanted to be able to setup internal services, and then on the go be able to access them. While I could setup a L2TP or SSL VPN and connect whenever I wanted to use these services, I thought I would give On-Demand VPN via a iOS/macOS configuration a try. Little did I know the world of hurt I was entering. I will start with the settings you need to get it working, since a lot of people just want that. Then I will talk about the crazy and painful road I went down before finding 1, just 1, set of settings that seem to work. If you have any questions, thoughts, or success stories please comment below!
Fun fact: I will be calling the protocol IPsec here. That is what the original RFC called it, what the original working group was called, and the capitalization they used. Sophos agrees and uses that capitalization, while Cisco and depending on which web page you are on for Microsoft may call it IPSEC or IPSec or IPsec.
On-Demand VPN gives you the ability to set certain websites or IPs, and when your phone or laptop attempts to connect, the machine silently brings a IPsec tunnel online and uses it for that traffic. This allows you to run services at home, and to users (your mom or cat or whomever) it looks like just another website. Apple has 1 big requirement for them, you have to use certificate based auth. You can not use a pre-shared key/password. Also up front, to save you a few days of trying things. iOS and macOS will NOT check your certificate store for your VPN endpoint (Sophos XG) certificate, it HAS to ship with the firmware or you will get the fantastic and descriptive “Could not validate the server certificate.” Also believe it or not, that is one of the most descriptive errors you will get here. There are some posts on the Apple support forums from Apple engineers saying the root CA has to be in already on the device. If anyone gets it to work with your own let me know.
Sophos XG Setup
I am using Sophos XG v18 with a Home license, backed by AD running on a Dell Optiplex for this guide (dont worry it as a cool Intel Nic in it). To setup the IPsec server in Sophos XG first we need to make 2 certificates. Login to the admin portal, then on the bottom left select “Certificates”. You need 2 certificates; 1 is our “local certificate” (we will call it Cert-A) this is a cert that is used for the server (Sophos) end. As previously mentioned, this has to be a real signed cert. I ended up forwarding a subdomain on my site to the firewall, and then using Let’s Encrypt to create a cert for that URL. I used this site, https://hometechhacker.com/letsencrypt-certificate-dns-verification-noip/ to guide me in creating the cert on my laptop, then I uploaded that to the Sophos firewall. This will require you to have access to your domains DNS settings or be able to host a web file.
The second cert (Cert-B) is for the client, Sophos will call it “The Remote Cert”; this is to auth to the firewall, that can just be a locally generated cert. All devices will share this cert. The devices will use their username and password combination to identify the user. I used email as the cert ID, note this email does not have to exist, I just made one up on my domain so I will know what this cert is. Once created, go back to the main Certificates page and download the client/remote certificate, I suggest putting an encryption password on it since the Apple tools seem to freak out if that is missing. But ALSO the password for this cert will be in clear text in your config, so don’t make it a password you care about. These certs all need to be rotated at least once a year, with the newer requirements; Let’s Encrypt is every 90 days and I intend on automating that on one of the Linux machines I have.
Now that we have our 2 certificates, lets go over to “VPN” on the left hand navigation. I have tried many settings in the main “IPsec Connections,” and none of them have worked for me. I get fun and generic errors from the Mac of “received IKE message with invalid SPI (759004) from other side” or “PeerInvalidSyntax: Failed to process IKE SA Init packet (connect)”.
Click the “Sophos Connect Client” tab, the back end of this client is just a well setup IPsec connection. Fill in the form, from the external interface you want to use, to selecting “Digital certificate” as your auth method, followed by the “Local certificate” which is the Let’s Encrypt one (Cert-A). “Remote certificate” is the one we will load on your device (Cert-B).
Now you select which users you want to have access to use this. I have Active Directory backing my system, so I can select the AD users who have logged in before to the User Portal. This is a trick to Sophos XG you may need, if you use AD and a user doesn’t show up, that means they need to login to the User Portal first.
Select an IP range to give these clients, I suggest something outside any of your normal ranges, then you can set the firewall rules and know no other systems are getting caught in them. Once you are happy, or fill in other settings you want like DNS servers, click “Apply”. After a second it will activate, you can download the Windows and Mac client here, or follow along to make a profile.
To create a configuration file you need to download Apple Configurator 2, https://apps.apple.com/us/app/apple-configurator-2/id1037126344 onto a Mac. I know what you are thinking, 2.1 Stars, Apple must love enterprises. Download that from the store and open it up. If you do not have a Mac I attached a templatethat you can edit as a text document down below. This profile needs a Name, as well as an identifier. The identifier is used to track this config uniquely, if you update the profile, then your device will override old configs instead of merging. You will see on the left there are LOTS of options you can set, the only 2 week need are “Certificates” and “VPN”.
Starting with Certificates, click into that section, then hit the Plus in the top right. Upload the cert we exported from Sophos (Cert-B) earlier for the end device, and enter the password for it. Again note, this password is in plain text in the config file.
Now for the VPN Section. Click the Plus in the top right again to make a new profile, name the connection anything. Set the Connection Type to “IPsec”. IKEv2 is IPsec but a newer version, I will get into some of this later after our config is done and I can rant. Server is your Sophos XG URL. Account and password can be entered here to ease setup, or you can leave one or both blank to make the user enter it when they import the config. You can leave the user/password fields blank (it will give you a yellow triangle but that is fine) and then give it out widely and not have your creds in it… For “Machine Authentication” you want “Certificate”; you will see in selecting “Certificate” all of a sudden the On-Demand area appears. For “Identity Certificate” select the one we uploaded before. Finally we can enable “Enable VPN On Demand” and select the IPs or URLs you want to trigger the VPN.
Once that is done, save the profile and open it on a Mac or you can use this configuration tool to upload it to an iOS device. That should be it! Your devices should be able to start the connection if you ask it, and if you go to the website should auto vpn. Make sure you have firewall rules in Sophos XG for this new IP range, or that can block you from being able to access things.
A small note, from my tinkerings with the On Demand profile if you go to Safari on a iOS device, it will connect when you visit a website that is in the configuration. If you use a random app, such as an SSH application, I didn’t find it always bringing the tunnel up, and at times it had to manually be started. Something to lookout for, a nice part of the the IPsec tunnel is that it starts quickly.
Now that the config is done, I want to mention some of the other things I have learned in tinkering with this for several days. The only way I got it to work is using that Sophos Connect area, and the other big not documented thing is you have to use a publicly trusted cert for the Sophos end. I found 1 Apple engineer mention this on their forum, and a TON of people talking about how they couldn’t get the tunnel to work with their private CA. I have tried uploading a CA, and injecting it different places with different privileges for the Mac and never could get it to work. The Let’s Encrypt cert imminently worked.
For IPsec v1, aka IKEv1, Apple uses the BSD program racoon on the backend to manage the connection. Using the “Console” app you can find the logs of this. For IKEv2 it seems Apple wrote their own client around 2016-2018, there are a lot of reports online that it just doesnt work at all with cert based auth. All the guides about it working stop around 2016. You can find earlier ones, or people using pre-shared keys, but selecting pre-shared keys doesnt allow us to do a On Demand VPN. The bug has been reported for a while, https://github.com/lionheart/openradar-mirror/issues/6082. If you try to do this, you can expect A LOT of “An unexpected error has occurred” from the VPN client. Even looking at the Wireshark traffic didn’t lend any help on tuning Sophos to give the IKEv2 client something it would accept. If someone figures out how to get that to work in this setup please let me now.
Now that everything is setup you can host things yourself. I give the auto connecting VPN less rights than when I do a full tunnel on my laptop, but it allows for things like Jira to be hosted, then mobile clients to easily connect.
For your cert to work in the template it needs to be converted. Sophos will give you a .p12 file for your cert, use the following command to get the version that needs to be in the .mobileconfig file. You’ll at minimum want to edit the cert area and put yours in there, set the password for the cert, and any URLs you need.
The Mac journey continues with me searching for a way to transfer files from my modern PC/Mac onto the old Macintosh SE I recently was restoring; a way without constantly removing the SD card from the SCSI2SD adapter and mounting it in an emulator. After reading a lot of different pages, and hitting different dead ends, or methods that involved a lot of hardware, time, or monetary investment I found an old reliable way to transfer files.
One of the methods I looked at was an ethernet LAN adapter for the Mac SE; the issue I saw was some of them were expensive and a lot of them required more RAM than the 1MB my SE had. I then turned to the serial ports available in the back of the machine. The Mac does not come with a lot of software to help in this endeavor, which made me use the SCSI2SD adapter to load the initial setup on, then I could use the software to transfer after that.
I ended up using the Kermit protocol, the same protocol used to transfer software to the Compaq Portable II. The project was run by Colombia University for many years. While they have since transferred it to be an open source project, the original project files are still on their FTP server, and this offers everything from DOS to Mac to C64 binaries. ftp://columbia.edu/kermit hosts all the files, for archival purposes I also uploaded a clone of that folder to archive.org; https://archive.org/details/kermit_202008 . Kermit is not fast, being serial and the Mac can’t support anything over 57600 baud; but it offers compatibility with almost every OS at this point. Get ready to experience what dialup was like all over again.
Serial Adapter for the modern computer if your system doesnt have one on it
To start the connection, I will be using a modern Mac as the server (a modern Mac being a 2012 Macbook Air), and a USB Serial cable to connect to the Mac SE as client. Using homebrew on the Mac, you can install “c-kermit”. Once that is installed search for your serial device under /dev/, mine is /dev/tty.usbserial1420. Please note wherever you start kermit, will be the home folder for file transfers, I suggest making a folder somewhere that you will drop files to transfer.
> set port /dev/tty.usbserial1420
> set carrier-watch off # Assume there is no carrier signal
> set speed 57600 # Or whatever the speed has to be
Get ftp://columbia.edu/kermit/mac/mackermit.hqx and get it onto your Mac SE, through some means. I transferred the whole “mac” folder from Colombia’s FTP server onto my Mac SE. I would suggest a SCSI2SD adapter for this initial transfer. You may be able to use a floppy, but you may hit issues depending on your model of SE. Mine has a 800kb floppy drive, so results of writing floppies from a modern PC usually end with it not reading them. Modern floppy drives are cheap working at 1.44mb, and the tracks wont align. Once you have the Kermit app on the Mac open it up.
Select “Settings”, at the top, then “Communications”. Here you can set the speed to the max speed supported of 57600 over the default 9600 baud. Both of these are terribly slow… but there is nothing we can do about that. Make sure to select the Phone or Serial port based on which you are using; I used the Phone port.
Afterwards, click the “File-Transfer” menu at the top, then “Set Directory” to set where the files transferred should end up. Then open the same “File-Transfer” menu again and “Get file from server”; here you can type in a filename that exists in the folder you opened Kermit on the Server.
Now be prepared to wait for a while… Eventually the files will be in the folder you selected and you are good to go!
A few things to look out for, if you have a older Mac SE like the one here and it only has 1MB of RAM, that means you can only run Mac OS 6. (https://www.lowendmac.com/oldmac/compact3.html) I may upgrade this system in the future to its max which I believe is 4MB, but for now I am stuck with 6. This also means I can only use DiskCopy 4.2, and some good amount of classic apps will not work on Mac OS 6. The biggest issue is there are a lot of archives that are in DiskCopy 6 format, which I can’t load on the system.
The first thing I thought I would do is extract the archive on an old Mac VM on my modern computer, then transfer the files onto the Mac Se. Here I ran into a lot of issues with the file types that exist. If you want to go down a weird rabbit hole, the classic Macs used an odd 4 letter system for the file type, and 4 letter for which program created it, http://livecode.byu.edu/helps/file-creatorcodes.php . The Mac mostly ignores file extensions. There are programs such as ResEdit (that comes on the provided SCSI2SD disk image I used in restoration) where you can edit these attributes, but it usually leads to weird outcomes. Kermit tends to bring files over as “text”. StuffIt seems to do a decent job of just looking at the file extension and allowing you to expand it, then those files are the correct type. This whole issue is something to look out for, doubly so on a System 6 machine and can not run DiskCopy 6.
Otherwise stick to websites that say they backup with DiskCopy 4, or get more RAM… Then have fun with the system! Write that novel you have always wanted to write without distraction.
Years ago someone gave me a Macintosh SE, 20MB SCSI HDD, with 1MB of RAM. I had it sitting in storage and decided it could use some new life; this involved what I found out to be repairing, upgrading, and getting parts for the little machine. Then I was able to come up with a modern way to transfer files to it, so I can get software off the web, then get it onto the system without too much hassle, but that is getting ahead of myself.
Last time I used the machine I remember it working, but then when I went to turn it on the system gave a sad mac with an error. In looking it up, http://www.midiguy.com/MGuy/MacQs/SadMac.html#anchorSE&II, I was told it was a RAM error. Power cycling the machine would periodically change the error, and once in a while get the machine to power up.
Getting the case off needs a special long screw driver, which I happen to have. The back only has 4 screws and then lifts off. Any repairs to these systems have to be done with a lot of care since there is a high voltage CRT. Very carefully I removed the cables from the motherboard, and then removed the motherboard itself.
After I removed the case and looked at the RAM, it was fairly oxidized. I happened to have a can of deoxite, I removed and cleaned all the ram and then the DIMMs. After, what I will say was jankily setting up the motherboard, it booted the first time. I did notice one of the legs on the the little slots didnt look at good as the rest, but it seems to work fine now.
Luckily for me the 20MB, 3.5″ SCSI drive still works fine. I ran diagnostics on it and them came back clean. I wanted to be able to download files from a more modern system by I will do a different post about that.
There were 2 more upgrades I wanted for this machine; first the original 1987 PRAM battery was still on the board. Fortunately it had not leaked at all, but I still want to remove it. I purchased a new 1/2AA, 3.6V battery holder and thought I could use the expansion slot in the back to hold it. I am not using the slot, and that way when the system goes into storage I can pop the battery out easily. I had recently gotten a new 3D printer (Ender v3 Pro), and made a mounting bracket. It needed to be mounted on the inside of the bracket because of the high of the battery holder, but it works well!
The last upgrade I wanted was some sort of mass storage. (Mass storage being anything over a floppy with a few MB) I do have a second Macintosh, I think its a Classic but I need to go get it. Someone gave me a Zip 100MB external SCSI drive, but to get that working you need at least Mac OS 6, with the driver installed. The Mac also only has a 800KB floppy drive, making it hard to transfer files to. I have a USB floppy drive, but these newer USB drives are fairly locked to 1.44MB floppies, as well as I couldnt easily read the file format for it.
Enter the SCSI2SD (v5.5 Pocket Edition)! I got it on ebay from https://www.ebay.com/itm/SCSI2SD-V5-5-Pocket-Edition/193496539667, I don’t know the seller, but the item is great. It allows you to write a disk image you make with Mini vMac or Basilisk II onto a micro SD card, then boot the Macintosh from it! Boom solid state drive for your Macintosh. This also allows you to kickstart the process of getting an OS and software you need to hook the Macintosh up to something more modern. There are different models of these SCSI2SD adapters and different versions. Apparently v6 is faster for some systems version of SCSI. My main feature I wanted was a DB-25 connector directly on it, since a lot of these adapters come with an internal header, and I wanted this to be able to go between Macs.
In researching I found this blog, https://www.savagetaylor.com/2018/01/05/setting-up-your-vintage-classic-68k-macintosh-using-a-scsi2sd-adapter/ it has a great guide on how to setup the device and even images to get you going! (I backed up a lot of the files related to the adapter on archive.org if years later anyone needs them) I’ll skip over that since that blog covers it so well. The device allows multiple SCSI device emulation. Note, if you have a Macintosh like mine that has an internal HDD, that is SCSI ID 0, so make your device 1 or later. When booting the Macintosh you can hold Command-Option-Shift-Delete-# to boot to that device. With this setup I was able to transfer an OS install onto the ZIP disk (at 100MB plenty of space), and update the internal system.
I installed Mac OS 6.0.8, later editions need more than 1MB of RAM. For anyone with a similar system I would suggest running in Finder mode, and not Multifinder. Multifinder kept running low on RAM when trying to run applications. At this point the system is up and running, reliably, and I was able to put some games on it. I will have another article about using our old reliable Kermit to transfer files to the Mac!
Different network gear I have has had many problems trying to get email alerts working. I thought I would document them. All of these systems use a service gmail address I made on free/public gmail to send alerts to me.
Sophos, and LibreNMS gave me no problems; if you have issues with them drop a comment below and I can post my settings.
The trick to getting Ruckus Unleashed, I used “smtp.gmail.com” and port 587. The issue I ran into is the service email I use to send emails had a long password. Ruckus Unleashed v200.8 supports a maximum of 32 character passwords. I would also mention it dumps the password raw into the logs, so make an account you dont care much about.
After digging through logs and getting lots of “There was an error sending the test email to firstname.lastname@example.org. Failed to send email for unknown reasons.”, I found one post that mentioned a fix for the console log of “fail to send email: api.err.SmtpSendFailed”. You need to once again use smtp.gmail.com, and port 587, but since its TLS, you need to counter intuitively UNCHECK “Enable SSL”.
This post will be a bit more brief than some of the others, I was relaxing around Thanksgiving and put this together. Only afterwards did I realized that I was having such a good time, that hadn’t taken too many photos.
The kit comes from Chris over at https://www.adwaterandstir.com/altair/. The version I have is The Altair-Duino v1.4, which came in a bamboo box. There are now other versions, some with acrylic cases! This post will be about version 1.4.
The kit comes with all the parts you need inside the box. The main controller is an Arduino, hence the name The Altair-Duino. There is an SD card that you bend the prongs on (more on that later) which holds the disk images. This is a fun straight forward kit, that comes with everything you need minus solder. The Arduino came with the firmware it needed, and the SD card came with disk images preloaded onto it.
The kit comes with a spiral notebook of instructions on how to put it together. These are great, color photos of step by step what to do. You can see them here, https://www.adwaterandstir.com/instructions-14/ , keep in mind this is for my specific version. Like many of the other kits, the longest part of this kit is soldering all the LEDs and resistors onto the board. There are a few ribbon cables that go into place, and you are set. Be slightly careful when putting the switches in, they can be a tighter fit into the holes which is great for stability, but they are at the center of the board and it can flex. Once you get it all in the case and screwed down, clearance is a bit low, so make sure the board is ready to go in, when you put it in.
The one part of the setup that is a bit scary, the system comes with a SD card reader that sits flush with the board; if you want it to be accessible from the back of the case you need to bend the 4 legs on it. I used my trusty Radioshack wire stripper/pliers for that!
I connected over USB, the kit also supports Bluetooth on Windows, to get the serial line out and console in. The system supports loading a bunch of programs that are included. The creators website, https://www.adwaterandstir.com/operation/ includes a bunch of guides on things to do. I loaded up CP/M and for fun, of course Zork!
A easy kit to put together, and a fun little project. I now am amassing a wall of these projects, and will have to get a new shelf for this one. Then I will just wonder where Chris found 256mb micro SD cards!
TL;DR; Check that your Domain Controllers are in the correct OU and that Microsoft Key Distribution Service is running
I ran into an issue recently when DNSSEC signing a dns zone where Windows Server 2019 gave a very vague error, and would only display that error after 10 minutes of timeout. This made iterating on it very slow since every change I made was a 10 minute wait. Every guide to setup DNSSEC mentioned right clicking the zone, then clicking sign and as long as you select the default it should just work. On another domain, that happened for me and it just worked; except the one original one that kept timing out.
In setting a custom DNSSEC signing policy I noticed that there were different keystores each of which gave a different error. This made me think it was something to do with the specific one I was using. It was time to troubleshoot the service itself not DNSSEC.
I got a list of the services from a known good, and signing, domain controller; then compared that to the bad one to see what was different. Part way down the list I noticed that Microsoft Key Distribution Service was failing to start, and if I tried to start it, there was an error.
Group Key Distribution Service cannot connect to the domain controller on local host Status 0x80070020.
Checking the Event Log showed an issue in finding the Domain Controllers on the network (error above), which was weird because it is a Domain Controller… In looking at where this system was placed in the domain tree, I saw it had been moved from the original OU for domain controllers to another place. I dragged it back, after applying all the GPOs that were on that other folder to the original Domain Controller folder. Then held my breath, hit start on the Key Distribution Service and it started right away.
Setting up AD auth in the product is straight forward, set your domain search as wide as you are comfortable with, because next you import groups that are under that search. Next, make sure to hit the little icon that imports all the AD groups you want, it is easy to overlook.
Now go to the Services tab, and include your new AD servers in your group for Admin Authentication methods. The guides say to make AD first, and in testing I just put one of the servers above local; but this shouldn’t matter too much, local auth still works.
Now here is the trick that got me. TO HAVE THE USER SHOW UP IN THE USER AREA OF AUTHENTICATION, YOU MUST HAVE THEM LOGIN TO THE USER PORTAL FIRST. Thus the User Portal needs to also be setup to allow AD auth. After that, the user will appear like below, and you can click in to edit them.
Clicking into the user you can make them an Admin, and set their group. You have to provide a email at this point for the user. BEWARE, MAKING THE USER AN ADMIN IS NOT REVERSIBLE! IF YOU WANT TO MAKE THEM A NORMAL ACCOUNT AGAIN YOU NEED TO DELETE THE USER, AND IF THIS USER IS USED IN ANY FIREWALL RULE OR SETTINGS THIS WILL BE BLOCKED UNTIL THEY ARE REMOVED FROM ALL OF THEM. One fix for this is to make them part of a Admin group that has no rights to anything, but that doesn’t feel like the proper way.
Then you should be good to go!
Some troubleshooting techniques I used while fixing this: if you don’t have the user imported into Sophos XG, and attempt to login to the Admin panel, you will get “Wrong username/password” and looking at the logs in Sophos you will see “Wrong credentials entered for x@domain”. This is not exactly true and can throw you off. If you login to AD and look at your servers Security logs, it says “User login successful”. That is a good indicator that at least your login is working correctly, don’t get fooled by AD saying success, while Sophos says wrong; the user just needs to login to the User panel first to link the accounts.
In my apartment I needed to get wired networking with VLANs across the apartment. I didn’t want to run a wire since I thought my roommate would not appreciate that. I wanted to have a switch near my desk, that allowed different devices I have like file server, desktop, and a few other things to have a wired link; then, connect to the modem/firewall and rest of the networking gear across the apartment.
Long story short, I ended up using a trick I didn’t know would work till I tried it. I have 2 x UAP-AC-M, they work decently well, topping out at 867Mbps and 2×2 MIMO; as well as being able to get them on sale in a 2 pack for a decent price made them a great deal. I have run 1 of them for 4 years as my main access point. Then when I wanted to get this wire connection in a new room configuration I tried to do a wireless uplink to the second one. This makes it mesh with the first access point. Now the important item I don’t seem written anywhere but works well (caveats below):
Ubiquiti access points in wireless uplink/mesh will bridge that network to the wired port on the device
This means if you have a trunk port going into your original/base mesh AP, you will have the same trunk port coming out the other end. This also means anyone who is running mesh points, and hasn’t secured the wired port may want to think about doing so. I am will skip over HOW to set this up, Ubiquiti has a good guide https://help.ui.com/hc/en-us/articles/115002262328 to walk you through it, and most APs can do wireless uplink at this point; this is more about saying it can be done, and works well from my experience to anyone thinking about implementing this or wants a solution for their home/apartment that is not powerline networking. The APs I have are 2×2 802.11AC, I’m sure with a 4×4 AP like the AC-Pro as your base you may see better performance on higher trafficked lines.
This setup has worked well for me for over 6 months now, I can easily hit the 300Mbps I get from my internet connection on a desktop plugged into this meshed AP’s port; I also get 6ms pings to servers while playing games. You get the benefit of real commercial grade antennas and radios in the APs you are using instead of a tiny wifi chip in a laptop, desktop, or device. This also lowers the number of wireless devices (since all the wired devices would have been wireless instead). I also disabled the secondary AP from hosting any of the SSIDs I have in the apartment, so it just works as a wireless uplink. My apartment is not big enough for 2 AP’s for devices.
I am looking to move away from this setup for a few reasons. It has worked well and if you are in a pinch I would recommend this setup much more than powerline networking which I have also tried and used several times. I am hoping to move to 10gb/s networking at home with my growing homelab setup; thus, no more wireless link. The other limitation that 99% of people probably would not care about is that you can not do jumbo packets over wireless, so that means it can not be done from all I have read over a wireless link of this type.
The first caveat is that this configuration slightly confuses the access point when it first starts up. The first 60 seconds or so when the access point is online it will think the wired connection is its uplink and attempt to ping out over it. After that it realizes it cant hit anything and will go to wireless uplinking. Sometimes everything just works then, sometimes I have had my switch be confused about where traffic should go and had to power cycle it; in this case it was just a Netgear Prosafe switch with VLANs, not especially smart, but not the dumbest switch. This is similar to a enterprise networks re-converge time when a link is downed. Overall it is rarely a problem and these APs are solid and can go months between restarts, but this is something to lookout for.
Remember that if a Ubiquiti AP cant get an IP, then it doesn’t broadcast SSIDs; this is important since if the base AP boots (like after a power outage) and doesn’t get a DHCP address quick enough, it wont broadcast, then the mesh side will never find an uplink to connect to.
With the earlier mentioned topology issues you can run into, that can make management difficult. You need to make sure the base side of the network is stable. You can get into a position where you did a bad config push or a setting is wrong on the secondary/mesh side and the only way to fix the config is bringing that AP back to the original wired network and pushing a config to it, before the secondary AP can go back into wireless uplink mode.
One of the projects I currently work on at work, and have for the last few years is how to go from a blank stack of servers to a fully configured cluster with my companies software running on it. While some projects were starting and getting going in the open source field when I started this project 5+ years ago, a lot of them kept rewriting their API every minor version rev. That started my down a path that has now become a decently large internal network booting infrastructure, and managing interconnects to our inventory system as well as other systems such as Tenable Nessus. I recently was awarded my first patent! This one is specifically about how my system interacts with the inventory to dynamically assign systems as they come online to clusters.
My part of the code was all written in Java and continues to evolve as a platform, I hope to open source a good amount of it down the road. I started the project by reading the RFCs for DHCP/PXE and then writing code. I have grown to enjoy writing libraries and some project this way of adhering to the standard (more on that some other time). The general platform can handle ProxyDHCP PXE booting, and then uses iPXE to create menus and boot systems. I spent many hours debugging different vendors PXE code and BIOS vs UEFI to get all the systems to work. The platform now supports plugins for many different aspects of server configuration.
I could write page about small details I have learned a long the way; one issue that has been driving me crazy recently, if you want to ProxyDHCP instead of using your main DHCP stacks these days is Secure Boot. iPXE does not have a Secure Boot signed image, I have tried to get Microsoft to sign it but they will not unless you are selling a product using that the sign iPXE. I am not I just wanted it for internal use. That means you may want to use grub2 as your loader, but there is a bug that has been outstanding for over 6 years and makes ProxyDHCP with grub basically impossible, https://savannah.gnu.org/bugs/?55636 which is sad.