Troubleshooting the Centroid Acorn and Ether1616 Network | Why won't it connect? | Beelink + Realtek
Вставка
- Опубліковано 6 жов 2024
- Today we're troubleshooting why the Beelink computer cannot see the Centroid Ether1616 I/O board in my new CNC controller build. It turned out to be a really subtle (fun) issue. The system is running Centroid CNC12, with an Acorn and Ether1616 I/O Expander. And when we're done, it will all be working.
This appears to be a bug in the Realtek GbE network hardware of the Beelink computer. But there's an easy workaround.
00:22 Intro
01:15 Demonstration
02:53 Extracting the board
04:43 Checking voltages and PINGing
07:35 Sniffing the debug port
13:38 Testing with another computer
15:47 Testing with another network card
17:11 Sniffing the network traffic
18:55 A brief lesson in ARP
21:52 Analyzing the network traces
24:24 Network driver settings
26:33 Fixed!
Tools used in this video:
*This site contains affiliate links for which I may be compensated
EEVBLOG Brymen BM235 Multimeter (Amazon*): amzn.to/2YKFSEk
Saleae Logic Pro 16 Logic Analyzer (Amazon*): amzn.to/3h3kaXN
Internetworking with TCP/IP Volume One (Amazon*): amzn.to/3p4uMKD
Beelink SEI8 Mini PC (Amazon*): amzn.to/35kOMBK
Portable 14" Touch-Screen Monitor (Amazon*): amzn.to/3BCczJg
USB C to Ethernet and USB Hub (Amazon*): amzn.to/3BH4JOM
High Temperature ESD Safe Mat (Amazon*): amzn.to/2Xz0n6D
Raw Kevin MacLeod (incompetech.com)
Licensed under Creative Commons: By Attribution 3.0 License
creativecommons...
"if you don't know what this is, it's called a book. This was used in the before times to encode information on slices of dead tree"
This totally got me
It's seriously shitty legacy tech. Have you tried doing a search through it...?
My computer science lecturer used to (and apparently still does) introduce the concept behind a binary search by ripping a phone book in half until he finds someone’s grandma’s number… then called it in front of the lecture theatre!
It's fine when you've only got to search one book, they create indexes at the back for that... not as good as google but surprisingly effective for the common search terms. It's when you have to search for information in archival book repositories it's a problem.
These things called Libraries you see... except rather than pointing you to a single book the search index for Libraries barely points you to the correct shelf and sometimes people fail to put the books back in the right place anyway.
Checking voltages. Not only would you hear Dave Jones ' voice but you'll also hear Louis Rossmann yelling: "did you check PP3V3_G3_HOT" ? :-)
PP buss. I never opened a MacBook once, but that's lodged deep in my head.
Having worked in networking for the past 20+ years, as soon as James mentioned that this board was using Ethernet and that there were communication issues, I was suspecting that there was a Realtek chip involved. These are very cheap Ethernet chips often found in low-cost consumer hardware and they always seem to have weird issues if you are doing any type of networking more complicated than simple web browsing, and even then they tend to have weird latency and jitter issues. This is why Intel chipsets are the defacto standard for 1G networking (including 10/100 speeds too). Manufacturers like Realtek Ethernet chips due to their cost, users are better off with anything but Realtek, due to the shortcomings inherent to Realtek chips.
Same for me, in early days it was always a Novell 10-base 10 card that worked and the ne200(0) cards that failed in odd ways, now it's realtek that drives me nuts with linux issues where my intel based units work fine.
While I (sysadmin here) agree that the Realtek card are not always stable, and sometimes have weird implementations. The arp broadcast functionality is one of the very base block oh the tcp network stack. I doubt that even them would fail this spectacularly on a simple mask calculation. I’m wondering if the implementation of the Realtek chip on the nuc board itself isn’t at fault. Or some hardware init’ that should have been handled by the bios on boot that wouldn’t happen ? Those Chinese nuc (I used a few for dynamic displays before replacing them with pi’s) sometimes have basic functionality mangled.
@@amahashadow It's more than that with Realtek chips. The don't handle jumbo frames, ARP, 802.1q, udp, promiscuous mode, any offloading (tcp, udp, arp, crc, etc.), multicast, ipv6 anycast, and more.
@@amahashadow I would say in this case that the centroid software sends out a packet that LOOKS like an arp packet, and got flagged as such by the offload processor, but in actuality is not an ARP packet at all. Hence when you turn the offloading off, it ends up coming out the cable as the packet that was intended, rather than being groked by the offload mechanism on the card and munged. The cheap cost of realtek chipsets comes at a price, "rack of testing". There's a bug somewhere - the driver, the hardware? Realtek doesn't care as it works in 99% of applications - good enough, ship it!
It's not just their ethernet, it's also their USB products. They are truly awful!
You are a naturally talented teacher. Thanks for presenting everything in such clear terms. I would add that the electronic lead screw is one of the best projects I have ever found on the net in thirty years of searching.
Thanks!
I second this for sure! Really easy to follow your thinking to arrive at the solution
I loved the historical element of the "Book". Great explanation. I'm looking forward to see it work.
James, great example of troubleshooting technique, particularly as it shows how to approach a situation where you don't know what is wrong and aren't starting with a system that did work and now doesn't (most troubleshooting charts assume that something used to work and now doesn't).
Back in the neanderthal times before the interweb (early 1980's), when I was the founding VP of the Hershey Apple Core (still member #2, although a lot of things have changed in the group), I helped Member #1 troubleshoot his newly built Apple ][ clone since I dabbled in electronics (wires inside of evacuated glass envelopes and chunks of germanium inside little metal cans), and had the necessary equipment. We weren't able to use the troubleshooting charts, so resorted to a symbolic diagram of the circuitry (schematic?) and poked around with my oscilloscope. When we lost the clock signal, we found two unsoldered resistors, applied iron & solder and Voilá, everything was right with the world.
Many of the details of your adventure were a little too esoteric for me (I was OK up to and including hexadecimal, but then got glassy-eyed). However, the "book" thing really mystified me until I remembered a "motion picture" from the past called "ZARDOZ," where James Bond had discovered an entire building filled with items as you described, falling to dust as you would expect of dead wood. I now realize that these are the things that my parents had a vast collection of, which descended upon me and my siblings and are now stored in unused "guest" rooms.
All my best,
Charlie
I also appreciate the explination of 'how things work' in a layman's manner so even I can understand the process.
Thanks
this is probably one of my favorite videos you have made. I think many of your viewers aren't necessarily interested in a particular subject, but more widely in just learning new things. And as this was an area I happen to know a lot about (unlike machining, which is why I am here to learn) I was thrilled to see how you worked through the problem and used the opportunity to introduce folks to some networking concepts including ARP. Really really great. And it looks like you found a bug in that Realtek driver. Thanks so much for this great video. Would love to see more like this... just figuring things out and us along for the ride!!!
IMO this is the real point of difference for someone like James, and a 'dedicated' CNC (or whatever the topic you follow is), he can be agile, he can cover this weird networking bug today, then next week cut a swan out of a block of steel.
totally agree, I'm generally not interested in the main topics he covers (expensive hardware I will never own), his approach and methodologies is what I enjoy.
This went in a completely different direction than I would have thought. This mix of machining, design, electronics and now even networking stuff is what makes me love your channel.
I have never looked in detail to those network driver properties in Windows but this ARP Offload setting being the problem was a good find!
I'm a retired Network Engineer and have used WireShark for many years. I was around when all the tools you have been using in this video were not around. WireShark is probably the most valuable tool that I have used. Like all things network, finding the arp offload issue is amazing. Good job. My handle when working was Jedi...
Good job Jedi...
Yeah, I remember the days when we had to lug around the Compaq portable sniffer. I think it was a Compaq. Looked basically like the Dolch, but with light colored plastic.
When Compaq was bought by HP that was a sad day. Having been certified in Compaq and also HP, those older bricks were more trouble than they were worth. Wireshark made it so much easier to troubleshoot weird networking issues.
Stay safe...
I am not very knowledgeable about electronics, I just enjoy your explanations man. You rock!
Thank you for taking the time to explain each element. I find that troubleshooting definitely forces us to understand what is actually happening. As others have already stated this was a great mix of software and hardware interrogation. I appreciated seeing your logic and technique in the piece by piece approach to locating the cause.
I ran a CNC shop for 8 years (2000-2008). At that time, I had little interest and no time to understand how the all of the elements come together to make such a machine operate. I'm now in my mid 50's I am discovering a great interest understanding how all the pieces fit together. I find myself obsessed with learning all that I can, and working towards a reason to build something that provides me the sense of accomplishment.
I am very interested in learning, understanding and applying how test equipment works. Your logic analyzer seems like an interesting tool. I want to understand that tool and the oscilloscope!
James, I think you are a genius at troubleshooting. It doesn't matter if it is a CNC machine, a circuit board, a Cadillac car or a fire truck, you can always figure out what is wrong and how to fix it. I guess that is what you do for a living but you are great at it.
Wait, cadillac car or fire truck? Are there some videos on this channel somewhere I've never seen??
Very interesting issue. I was starting to wonder if the netmask was wrong since it was trying to ARP the broadcast address, but an implementation bug makes more sense.
The net mask is set to /24. I did check that.
Nice one James… I like how you demonstrated a “binary” cut on the results of your testing and observations. Followed the decision tree down to the last twig.
Long experience. :)
Pretty nice debugging 👍 Reminds me of a piece of network hardware I had once in my hands that repeatedly brought down the complete network. Found out that its broken IP stack replied "yes, this is me" to every ARP request on the network, regardless of the actual IP in the request 😅
That's really interesting. I've never had a reason to hate on Realtek until now. Sounds like I'm in good company.
I've seen that behavior from Unitronics PLCs. A lot of industrial "ethernet" is very broken.
Wow, was that setup to force a proprietary switch or point to point hookup only?
@@seanclark8452 No. It was an industrial RFID reader controller which could be used in any setup. But, to be fair, I think it was pre-production firmware. The issue got fixed after I reported it.
@@RoterFruchtZwerg Oh, that makes sense then. I'll hard code things like that in code under development to make testing a specific section easier. It wouldn't be hard to forget to remove a test mod, especially if there were a bunch and you just missed one.
My degree is in computer science. I do applications-level programming for my job, but my interest has always been systems; kernels and drivers, hardware/software interface, embedded systems, parallel, distributed, real-time computing, etc. I love seeing stuff like this
There seems to be more interest than I expected. Or rather...UA-cam has shown this video to more interested people than I expected. :)
@@Clough42 seems like everyone enjoys seeing a big being found
Wow! What a sneaky little bug.
Great demonstration of the fault finding process. I can't say I've ever seen that particular fault before, although I'm always suspicious of any power saving features on devices that misbehave, and disable any obviously relevant configuration settings. Bitter experience has taught me that there's usually something that doesn't quite go right upon resume, and sometimes on initialisation too. ARP Offload wouldn't necessarily be one of them at first sight, but now we know.
Thanks for taking us on the journey of discovery James.
This is an awsome troubleshooting session. I'd never been able to find a simiar bug myself, but seeing you doing so - my compliments!
There's something really strange going on here - ARP offload is supposed to let the NIC RESPOND to ARP packets while asleep, not interfere with the OS's ARP requests. There's clearly a bug somewhere but that's not at all where I would have expected it. Maybe Realtek's ARP Offload means something different?
The problem here is that the hardware implementation of the ARP protocol was not correct. Instead of determining if the destination IPv4 address is a broadcast address, the packet wasn't transmitted. Instead an ARP request was generated, then when no response to an invalidly formed ARP request was received, the packet was dropped from the internal buffer on the Realtek chipset.
This implies that the NIC is doing both ARP requests and ARP replies, which, if that was the case, the ARP request wouldn't even show up in wireshark.
James I have been a user of the Acorn Centroid product for a number of years and the company has an active user forum as well as tech support via email for troubleshooting. I am surprised you were not able to resolve the issue with what is available and were compelled to go into such detail to resolve the issue. However, I am sure they will appreciate your feedback on the problem.
I opened a thread in the forum and was immediately assisted by a couple of people. After going through the usual suggestions to turn off IPv6, firewalls, virus scanners, etc. they were stumped. I reported my solution back to the thread and added a link to the solution in the first post.
Amazing - thanks James for a small glimpse into a world I know virtually nothing about….. yet I had no issue following along. 👍👍😎👍👍
As others have noted before, and below, your ability to connect the dots in such an effortlessly understandable manner is SO appreciated. Carry on, sir.
Great Video!
Curious if manually adding the MAC address to the ARP cache would have solved the problem (i.e., arp -s ).
Protocol handling in the logic analyzer was really neat to see in your video. I sort of knew it could be done but the last time I used one, I don't think those feature were common. At least not on the entry level equipment I was using.
One of the things I love about your videos is your ability to deep dive into any category along with your willingness to share the journey as your learn and figure it out. You are also one of the few people I know that thinks sewing machines are cool too. After all, it's a tool!!! THANK YOU.
In that case you add it to the software cache. The would problem was that the software stack was not used at all. Instead a broken implementation on the ASIC was used. So your suggestion would probably not change that.
Could still try though
Interesting idea. I did notice that when it is working, the broadcast adress does indeed end up in the cache.
Very nicely done with a 'surprise ending' as frosting on the cake! Although I have a four channel scope with decode functions, adequate for most data com issues, your utilization/ demonstration of the Saleae analyzer is going to cost me some money! Keep up the great work!
Well, that was a trip. When I saw the logic analyzer come out, I didn't expect this outcome...
Absolutely fascinating! Your "Mini-Snars" are always so easy to follow! BTW, you can use that descriptor if you like!😁
"...record information on slices of dead tree". Absolutely brilliant!! Priceless!!!
Not sure this video is juicy for the algorithm, but your viewer base certainly appreciates it. As always, learned something new from it.
Yeah, I'm a little shocked at the caliber of comments and questions I'm getting. I was totally expecting to hear "Less talk, moar chips!"
That was an excellent step through of a tricky networking issue and a very good explanation of ARP and TCP/IP!
I've been doing IT for over 20 years and that was the most clear and concise explanation I've ever seen, well done!
It's amazing how much I enjoyed a video that was mostly in Greek... but troubleshooting is, to a point, troubleshooting and I do understand that language.
Troubleshooting involves a pretty universal set of tools and approaches to start with the broad picture and continuously narrow down the investigation until all that is left is the troublemaker.
You are a good troubleshooter!
Thanks for sharing! :)
Great video. Its great to have a simple explanation of basic internet protocol. Its great after almost 81 years to learn something new.
I understood 2% of the words said in this video but watched it in its entirety. Nice content, James
Excellent video James! It has given me the confidence to jump down the debugging rabbit hole if I need to. 👍🏻
Little glitches like that can be so frustrating. Good that you were able to figure out the issue.
Loved all the detail as you chased down the bug. Brings back memories of my former life as a network architect.
That was some great and interesting troubleshooting. Definitely shows to be systematic and don't assume anything.
An odd bug that could certainly manifest some really odd behavior. There is absolutely no reason it shouldn't know the broadcast address regardless of IP or mask. This is also the exact reason as a professional LAN engineer I continue to stress to our desktop guys that we need to be buying laptop and desktop machines with proper network chipsets in them. The wireless chipsets can be even more wonky, and embedded SCADA systems are even worse. At one point back in the 90s we managed to get a small number 3com NICs that all had the same MAC burnt into them. That was fun LoL. Enjoyed the whole video but the logic analyzer was my favorite, never had the opportunity to use one.
Your explanation of what is a very complex problem is simply amazing.
Excellent video; many thanks for sharing! I'm just starting my second Acorn build, and this is just the sort of thing that would have caught me out if I'd been adding the 1616 board (not needed on the current machine, but may be required on the next build).
I just randomly found your channel and wow I'm glad I did. Your professional troubleshooting knowledge is top notch. I will have to dive into some of your other videos. All the best with this cnc project going forward.
You clearly know what you're doing so I'm sure the first thing you tried was blowing into the connectors.
18:90 LOL brilliant! Also, Dave Jones has been waiting a decade for a proper diagnostic video opportunity. He needs to get out of his comfort zone and try fabricobling some cnc stuff. So many opportunities! Well done, sir.
I am totally impressed with your debug capability, considering you are mainly a cnc machine shop guru. Very good detective work. I wish I had your patience.
The machine shop is a hobby.
@@Clough42 which is even more impressive!
Your network debugging skills are pretty good. "it's called a book. This was used in the before times ... " -- luvit!! And yeah, 'stupid stuff' like that can cost enormous amounts of time, which is why for many -- just get the one which some other engineers already qualified and follow the cookbook. Whenever you stray off the recommended one, you are going on an adventure, and most DIYers are not going to have anywhere near your network skills. Even so, I appreciate the many comments below from experienced network troubleshooters.
Fantastic as always James! Will need to explore using the Ether1616 now… I am not familiar with this device to expand my AcornSide note… and maybe you have this planned… but I’d be interested in a video on how you leverage Fusion with the imported electrical components to design your control panels.
“This is a book.” LOL! Good one, James. Great job of troubleshooting. 👍🏻👍🏻
I have 28 + years in IT. I watched the entire episode, and I understand your conclusion. I don't think it entirely a ARP issue. I think it is more of a WOL issue or adapter compatibility issue.
I have used several different small form factor pc's and pi's and have seen driver issues with several of them.
You fixed it but I would careful going forward, you have other driver going forward as Windows Continue to be updated on you NUC.
I smiled when you whipped WireShark on the problem, I've worked with it since it was called Ethereal. I laughed out loud when Doug Comer's book (a method of encoding information on slices of dead trees!) came into play.
WireShark was/is an essential tool in my late career as a network engineer for, among others, Cisco Systems. I used WireShark and tools like it (you can imagine my bosses reaction when, in 1990 I needed to order a Compaq "Lunchbox" and accessories for around thirty-five thousand dollars), if not daily, at least on a weekly basis for thirty years. Comer's book and also "Routing TCP" lived on my desktop.
Anyway, nice catch on the Realtek chip configuration. I'll store it away in my memory for future troubleshooting. ARPing for the broadcast IP was the key. I saw the high-values in the pcap and it made me wonder but I have no experience with that chip set.
It was nice to see all the steps in easy to follow order. Keep up the good work!
One man's agony is another person's joy have fun Thanks Frank for the video
Enjoyed the info. Retiring soon and will be starting my cnc router build. I'll watching...
As someone who works with computers and networking for a living, this video was AWESOME. Great example of systematic troubleshooting plus we get a weird cause as well. Thanks for the great video and you have my subscription :)
Very interesting stuff. I used wire shark one time just to find an IP address of a device I was working on and had no clue what I was looking at. Your explanation from that paper book thing really helped me understand more of what wireshark can do. Thank you.
I plan to purchase a system from Centroid and a as a cost saving measure I plan to purchase a non-Centroid mini pc. The detail is deep and hope I do not have the same problems. However, thanks to you a roadmap is in place.
You are a teacher. I’ve always been a mechanical guy at heart with family and friends in comp science I’ve always loved the idea of comp sci but never had a place to start so I ordered that book and that will be my place to start, back in the beginning of something. I’ve always cheated my way through programming buy forums copy and paste and calling friends. and I love building cncs. I need to know this stuff!!
I always learn something in your videos, but it is like a box of chocolates, you never know what you will learn.
Good teaching !!! My first time understands TCP IP
My knowledge of (And usual interest in) electronics could be written in large type on a very small piece of paper. That being said, sir, you are the only one I know who actually makes me think I understand what is happening. All others make me think they are speaking in Vulcan or something. Thank you!
Well I confess most of that went over my head but I did learn something and it was interesting to see you troubleshoot the problem. Thanks for sharing.
Very cool. Thanks for posting this. Always enjoy (other people's) debug sessions! :-)
Thought it was going to be a simple issue like a cable but as always you have impressed with going the extra distance to sort and show us too 🙂. Currently building a control box too, my first one and already learning a bit 🙂
A very accomplished beautiful logical troubleshoot.
i didn't fastforwarded any second of this video .. really good job 🙏
Brilliant problem solving! Quite impressive. I especially enjoyed the serial port R/E -- I've never seen that before.
Good work on tracking this one down. FYI, the x.x.x..255 would only be a broadcast address if the netmask is properly set to 24 bits. If your netmask is set to 16 bit, bit would be a perfectly reasonable thing to do to ARP for any address that ends with .255, but that is in the middle of the /16 as those are usable addresses. I'm sure you set your netmask correctly, but you didn't show it, so I figured I'd mention it.
Many people have pointed this out. The netmask is indeed /24.
I absolutely hate computers and anything to do with them, but I want to say I found your issue, explanation and repair absolutely fascinating !!!!!
Damn! This was like watching a murder mystery. *Lieutenant Columbo:* "One more question. There's something that bothers me."
Less than a minute into the video, and a life decision has been made. I’m never posting pictures of my control cabinet anywhere. 🤣
I lay things out and really try to account for everything, but wiring always seems to take up more space than I think it will. At which point I get frustrated and usually go with the “F*** it, if the lid shuts its fine” approach. This has the added bonus of making maintenance/repairs and troubleshooting incredibly fun, where I often explore the diversity of “F***” and occasionally invent new adjectives and phrases.
Nice. I know the feeling. This is not my first rodeo.
I would never think that something like ARP could mess up like this.
I would have started with WireShark because I don't have a serial analyzer. I recall the before times and encoding information on dead trees! I challenge you to do CNC over TCP IP over Avian Carriers! 🕊 Nice troubleshooting and thanks for sharing.
What, African ones or European? 😉
@@M_Northstar Any bird! I would accept waddling penguins. Hmm- Must be a bird, and >= 100 meters. Flight is optional.
Same. Wireshark is the first place I'd have gone after it pinged but didn't work. I'm not sure I'd have pulled the USB ethernet out and tried that though. That was a smart move.
@@feeseize9569 I have a few of those USB to Ethernet adapters but, with the ping and the Acorn connecting I would have assumed the NIC and driver were good. Once it worked on another computer I suspect I would have dug a bit deeper.
@@joeldriver381 this was such a weird edge case. ARP offload worked fine for specific IPs just failed (or rather shouldn't even have ARP'd) for the broadcast case.
I love the fact that you have the EEVBlog DMM. I have the newer one on my workbench....
Yeah, that looks really nice. Maybe someday.
@@Clough42 - it's a great meter, but I find its a bit overkill for most daily tasks. 99% of the time, I use my older Fluke. I only break out the 121GW when I need one of its advanced features, or if I need all my meters for a 4-meter set up.
When you brought the book out……that cracked me up.
Clearly, the best things to do would be to switch to LinuxCNC. Not clearly because of Acorn, or because of this issue, but clearly because if you did then you'd do your normal deep dive on it, share it, and I'd learn so much from it. So clearly, you need to use LinuxCNC. This might be a tiny bit selfish on my part. ;) Kidding aside, I always enjoy your videos!
I went with Acorn specifically because I didn't want to have to tinker. Guess that's going well. :)
Not hating on Linux as working on the platform has paid the bills for the last 15 years. However, in the case of LinuxCNC what's the betting that James will be re-compiling the kernel and the ethernet drivers for the 50th time while wearing a t-shirt that says same ? (jk) It's lucky there was an option in the device manager to turn that off otherwise things would have gotten really hairy.
@@Clough42 linuxcnc and mesa boards are rock solid. Have been using them for three years with no issues. I am a retired Cisco network engineer and a Microsoft Certified System Engineer since 1996. I went with Linux because I have first hand professional knowledge of Microsoft's "Best Practices"
@@AjaxCrypto you're not wrong. I installed what I thought was the right image for linuxcnc, ran the rtos latency tests, was underwhelmed and then discovered I was supposed to install a different image for the mesa card, and decided I didn't want the aggravation. I'm glad it's stable for you. I've been using Linux since I installed my first Slackware from floppies in the early nineties. It works great, but keeping up with the changes is exhausting, and I want to spend my brain cycles on something else. So far, that's not working out so well, but we'll see.
I'd never have found that! Probably a good general setting for "budget" boards to disable ARP Offload.
Great troubleshooting!! Checks all the boxes!!
Tell me more about this encoding information on the dead trees? /s
Great video love seeing these sorts of exploration of things that *just come up* IMO that sort of stuff is the reason to watch a more hobbyist channel over a Professional Machinist or Professional CNC UA-camr.
You have problems you explore those problems, heck sometimes you won't solve them, but we are coming along for the ride.
Great job troubleshooting! I would have gotten stuck at the ARP packets. I would not have realized that the broadcast query was unnecessary.
It took me a while to realize it was wrong.
Great job troubleshooting the problem. I am sure the most of us would never have found it. Keep on keeping on.
@18:55 "...if you don't know what this is, this is called a book." 🤣🤣🤣 Great video on troubleshooting....
I have no idea how you knew how to even begin that process let alone push ahead logically and find the solution... but it was awesome to watch.
Sad fact though is that if I want to do Acorn on my CNC build (and I do) that I'll likely run into some OTHER speed bump that will just end it for me. I mean I'm gonna try anyway, but I fully expect to be shot down before I reach the finish line. Not sure if that makes me courageous or stupid.
They sell preconfigured computers that they support. Like most problems that aren't caused by money, this one can be solved with money.
This is the kind of thing that sets my hair on fire. Good on you for having the perseverance to slug through this.
Just where do you think HIS hair went? He just didn't film that part.
I didn't see you mention your IP subnet mask. My initial suspicion was that the computer didn't know that .255 was a broadcast address because the subnet mask was wrong. I was a little surprised that disabling ARP offload fixed the problem. But, if it works, it works.
The subnet mask is correctly set to /24.
Very (to me) impressive troubleshooting👍👍👍
How do you find out about tools like Wireshark? Not to mention the logic analyzer. Your thorough understanding of the tools you have at your disposal, and watching you use them to find a foothold in a complex problem like this is really cool to see. Great stuff.
I hang out with smart people. I wrote low-level network management software for many years, and a sniffer is part of the standard toolkit for that kind of work. I have friends and coworkers who build embedded industrial systems for a living, so I get to see all kinds of very cool and very expensive equipment. Lots of that kind of gear is starting to filter down to the hobbyist market (Rigol and Siglent digital scopes, Saleae logic analyzers and a hundred similar devices, Der EE LCR meters, etc.)
I loved the Doug Comer book. I reviewed it once and gave an enthusiastic must buy recommendation and noted there was a bug in a sample implementation in it but it was a bug with zero effect. Less than two minutes later I got an email from Doug Comer saying, "What?" I explained the bug which was not following an rfc protocol exactly (but with no ill effect) and he got back to me quickly saying thanks. In the next edition it was fixed, but I didn't get any thanks in the book. Sigh. I could have been famous, lol!
I enjoyed the troubleshooting video. Thanks.
So many levels of wonderful. What an excellent troubleshooting exercise
I've spent many years working on a large scale IPTV platform which requires a very well implemented network stack to work properly. I can't tell you the amount of times I've run across poorly written hardware offloading on NICs (looking at you Broadcom).
Broadcom offloading has caused me much grief as well.
Great problem solution. You know you are in trouble when you have to resort to referencing one of Doug Comer's books to figure out what's not working!
Yeah. Especially volume three.
I do use wireshark from time to time for debugging but I never expected to see it in a youtube video 😂 I had never used a logic analyzer though. That is pretty neat and I'm going to have to get one.
Nice demo of packet communications and protocols. Drivers and how windows makes certain defaults can be equally frustrating. I have a test system that uses a USB bridge to Ethernet and is internal to the device. Windows and it’s fail sage registries (ya right) kept dumping my manually selected driver anytime I went from using a docking station to not. Even still using the same physical connection. Anyways, great demo!
I don't know why you don't have a million subs. That was a thorough and detailed account of how to troubleshoot. I use Wireshark from time to time and I was happy to see you use it. Also... this booook you spoke of, I have "a few", they are amazing. There's no boot time. They can't take the site down. Its cool tech. :)
Thanks and interesting. Good old wire shark. ARP offload typically the default for NICs, in my mind this suggests a bug with the non-communicative board NIC hardware/firmware rather than your NUC and I guess good luck with getting them to sort that out! as you suggest they would blame your computer. Anyway workaround succeeded - congrats
Wow, pretty obscure! Great debugging, *really* appreciate these debug videos!
Solid. I still have my 5 book Microsoft Windows TCP/IP networking manuals from around that time. Pity the trees. Good catch on the ARP bug.
I’ve done my fair share of network analysis. I would not have caught that!
Thanks this was great!
I am glad you found a workaround but Realtek NICs are widely known in other hobby/DIY circles (homebrew NAS and firewalls particularly) to be dodgy. You can't trust them and an Intel chipset NIC is the first thing I look at when shopping for motherboards. Note the vendor suggestion of a NUC.
Yup. This is why.
Brilliant episode the best yet I work on complex networks in industry and the transport world and would not have seen this issue as the equipment we deploy is generally high end stuff the NICs in the cheaper end are just that cheap I’m working on my own design and build CNC mill and will buy a 2nd hand quality PC over a new cheap one for the reasons you found here, keep up the great work.