so...I put a Time Server in my HomeLab

92.55k views6452 WordsCopy TextShare
NetworkChuck
Get NordVPN 2Y plan 4 months extra plus up to 20 GB Saily data voucher ➼ https://nordvpn.com/netwo...
Video Transcript:
I'm installing a time server in my studio. What does that mean? Well, typically to receive time we talk to another server somewhere else outside our network.
I don't want to do that anymore. I want to host it myself and talk to my own server. So when Time Beat sent me this, I knew it was time.
This is the open Time card and it's kind of crazy. It has a GPS module that receives signals from satellites with atomic clocks. Sitting on top of that GPS module is a Raspberry pi, CM four operating as a server.
It can sync all the clocks in my network to nanosecond level accuracy. Now in order to get nanosecond level accuracy from my network, we're not using NTP, which is normally how our computers will get time from other servers throughout the internet. No, the suckers using something I didn't even know about.
It's called PTP or Precision Time Protocol. Yeah, that's what this is a whole world. I spent a lot of time preparing for this video and learning about time.
Now, I'm actually pretty nervous for this video because I've not successfully set up this sink just yet. We're going to do it together. But before we get started, what do I mean by sync?
Sync up our time. Who needs that? We all do.
I mean, think about it. How does everyone in the world right now kind of have the same time? From Tokyo to Paris to Dallas, we all sync up our clocks somehow and we have the same time down to the millisecond.
Also, how do we even keep time? Let's talk about that because this is actually pretty crazy. In fact, here's your 62nd history of how we've been keeping time starting with the Egyptians in 1500 bc.
Let's go Sundials. Keeping time with shadows unless it's nighttime or cloudy. But hey, we can use Stars Trek in the rotation on 12 stars in the sky.
We can tell time it's sound familiar. In 400 BC we had water clocks, water flowed from one container to the other, measuring the passage of time. 13th century game changer, mechanical clocks driven by weights and measures.
We now had hours, minutes, and seconds. 1656 pendulum clocks. We can more accurately track time by the swinging of a pendulum at a steady rate.
This was our go-to for the next 300 years. Now at this point, there's still no synchronization of time between people. So in 1847, the British railways invented Greenwich meantime to make sure the trains were on time.
And then in 1884, the international primarily and conference made time synchronization global and invented time zones. Thank you for that. 1927 is fun though.
Quartz clocks. Did you know if you electrocute a quartz crystal, it'll vibrate 32,768 times consistently giving us an insanely accurate way to keep time so accurate. We still use that in our computers today with our RTC, but in 1967, time went nuclear atomic clocks.
We can use the vibration of CC M atoms and get nanosecond level time accuracy. That's a billion of a second. It's so accurate.
We outdid the earth seriously. Sometimes the earth will rotate slower or faster and our time is more accurate, so we have to adjust for his bad timekeeping, occasionally adding leap seconds to our own clocks. Atomic clocks can also use rubidium, but CM is more accurate.
In the 1970s, we strapped an atomic clock to a satellite and sent it to space. These GPS satellites orbit the earth and provide highly accurate time signals to everyone. We're revolutionizing how we navigate and synchronize time, but what about our computers?
Now, even though our computers had quartz crystals, helping them keep their time, they were drifting. My computer didn't have the same time as your computer or Bob's computer or Sarah's computer. Our computers had no way to synchronize your time across the world.
That's where NTP comes in or the network time protocol invented by David Mills. We can actually synchronize our time across the internet and get accuracy within one to 10 milliseconds. And this is what we use right now, but we can get more accurate and that's where PTP comes in.
We go from millisecond accuracy to nanosecond accuracy. Let's talk about that now before we talk about PTP or the Precision time protocol. Let's break down what this is.
This is the open time card mini from time B. Now, when they first sent this to me, I was like, what is that? That's why it sat in my closet for 18 months.
Sorry, time B, but now that I've had a chance to look at it, this is kind of amazing. Now, one of the coolest parts about it is right here on top we have the Raspberry PI CM four. This is the server part of the time server.
And if you've never seen one of these before, it's a compact version of the Raspberry PI four. It's specifically designed for industrial and embedded applications, and it's kind just a chip you slap onto a board like check it out. I can just pop it off.
And right underneath that, we have the secret to how this thing can get amazing, incredibly accurate time. It's A-G-P-S-G-N-S-S module for the Raspberry Pi. It's this thing that will receive signals from the GPS satellites that have atomic clocks on them and receive the incredibly accurate time.
And under that is the CM four PCIE IO board. This allows us to just plug this into any computer or server, a server inside a server. Now for testing, I'm just going to plug this into the side of my Zema board, which has A-P-C-I-E slot right on the side.
Actually, I need to power this up first, plug in the ZEMA board power. And by the way, it doesn't matter where you plug this in, as long as it's A-P-C-I-E slot that provides power. And before I plug it in, I need to add the antenna.
The GPS antenna. I mean I have it right here. Lemme show you what I bought.
I'm not sure what that accent was, what I was trying to do there. No idea. Oh yeah, it's this guy right here, bought it on Amazon's standard GPS antenna.
It's actually kind of big and magnetic, so I'm going to be careful that it doesn't touch anything on my computer desk. I did have to buy an adapter and SMA to TNC and we'll screw that bad boy in. Now I was chatting with Ian over a time beat.
He said, this thing ships plug and play. So we're going to plug it and play it right now. And what I mean by that is it already has a software pre-installed.
All I needed was the antenna and a power source. So I'm going to put this on the couch. It's kind of freaking me out.
Now, I will plug this into one of my legit servers when we're done playing here, but let's go ahead and power this guy on. Oh wait, no, not yet. I need ethernet.
What am I thinking? In fact, that's a crucial component of how PTP is going to work with our open timecard. It'll send specially crafted hardware timestamp packets over the ethernet port.
We're going to nerd out about that here in a bit. Let's get it plugged in. Fingers crossed for a blue light.
Yes, got it. Now because this is already pre-configured and I also did a bit of my own configuration, thank you Ian for the help. This is going to launch and with that antenna, it's going to start connecting to satellites and getting incredibly accurate time, nanosecond level accuracy.
That is something we can't even really comprehend, at least I can't. I'll try to explain it here in a bit. But while it's doing that, let's talk about where we're at right now with time, like me right now and you with our computers and how we get time.
How does that exactly work? Like I mentioned NTP or the network time protocol very briefly, but what's going on there and why do we even need something like this? Who needs this?
I do. I need it. Not really.
I have no need for this, but lemme show you why. It's very cool. Here is your computer.
How it keeps time is pretty fascinating. Now when it's powered on, it'll use the CPU and the OS to keep track of time, but when it's powered off, it'll use a quartz crystal. We just talked about that.
It's called the RTC or the real time clock. It might look something like this. This is not actual size.
I'm going to add an asterisk for that, just like serial. And yes, this is that same quartz crystal that when an electrical signal is applied to it, it will vibrate. Vibrate at a constant frequency.
37,268. Is that right? Nope.
32,768 times per second. Now you might be wondering, Chuck, if it's powered off, how is it getting power? It needs that charge.
Well, we use a battery. It's called the CMOs battery. You probably see it on your computer.
Probably wondering why does my computer have a watch battery to keep time? And I just got that. Oh my gosh.
It's a watch battery keeping time and your computer genius looks like this. And it's only used when your computer's powered off and it keeps that charge applied to that quartz crystal. So it always keeps time and it's very, very accurate down to the millisecond.
But we have a huge problem here. This is not good enough. There's a problem that plagues time.
It's called drift, not Tokyo Drft. This is the bad kind. That was a terrible joke.
Okay. Almost killed my time server. It was just a matter of time.
I'm going to draw drift in red to make it look scary Every day. This quartz crystal RTC clock will drift one to two seconds, meaning that if you set your computer's clock to an extremely accurate atomic clock ready set, it might be accurate. Right in that moment, a day goes by, that's off by one to two seconds, another day, one to two seconds.
You see where I'm going here? And it can be as high as 10 seconds. I'm just being generous here because things like temperature can affect it.
So if you give it enough time, you've drifted away out to sea and that's why you relate for lunch that day you thought it was dinner. Now why does this happen? I thought it was supposed to be accurate.
Yeah, but things like temperature can change that because remember, we're tracking the vibrations of crystals, the oscillations, the regular repeating motions, temperature can change how many times they move. The crystals themselves while cut to run at 32,768 times per second can have minor imperfections, aging power supply, electromagnetic interference, you name it, your clock can get off. So what do we do about it?
How does anyone ever know what time it's, this is where NTP comes in or the network time protocol. Lemme throw an NTP server out here. The NTP server is a time server similar to what we're doing here.
This is just way better. And actually I will be deploying an NTP server on this as well. Don't you worry.
Your computer will on the regular go. What time is it? Let me ask my NTP server.
He'll do this every 64 to 1024 seconds. It depends on your configuration. It can be shorter than this.
You can do like four seconds if you wanted to, but that's crazy, but not crazier than what I'm about to do here. Just saying anyways, so your computer will reach out to the server on UDP port 1, 2, 3, the easiest port to remember in the entire world asking what time it is and the server will respond with the time. It's hammer time.
Sorry. Now there are some synchronization messages that happen here, a bit of back and forth, but that's the gist. And when it's all said and done, your computer adjusts his clock and he has an accurate time.
But hold up, why trust this guy? What's this NTP server know that we don't, doesn't he have an RTC chip? How does he have better time than me?
Well, because he knows the guy who has extremely accurate time, we call him the atomic clock. You see this time server, this N TP server might be time nist. gov, and he's what's known as a stratum one NTP server.
He stratum one because he has a really good relationship with an atomic clock. It might be this one, the atomic clock operated by NIST in Boulder, Colorado. And this thing is nothing short of magic.
It's a C zm atomic clock and it's one of many throughout the world that has the most accurate time. And this server here time this. gov is a stratum one NTP server because he's talking directly to this atomic clock.
That's where he gets his information. He's not relying on his RTC chip who, no, no, he's talking to the big guy. Now there's more than one stratum.
You can go down to a Stratum two server. This would be a server that isn't talking directly to the source. He's not talking to an atomic clock for his time source, but he might be talking to a stratum one server.
Funnily enough, if you're a Windows user by default, our NCP server is time windows. com as far as I could tell. And research is a Stratum two server, which by the way, am I running that right now?
Lemme go check. Go to control panel date and time, click on internet time and yeah, sure enough time. windows.
com. I'm going to change that and we'll go with, it's sitting right there by default time. n.
gov. I want to get closer to the source, sorry, windows. Now stratum can go all the way to 15 and I don't know who in the world is connected to a Stratum 15 server.
Why? Because the further away you get from the original source of time, an atomic clock, the more latency is involved, the longer it takes for you to get accurate time. Now that begs the question, why does it matter if we have millisecond accurate time or why does it even matter?
We're a few minutes off for computers. It's a pretty stinking big deal. For example, funnily enough, as I was setting up this time server, I pulled it up, I logged into it, and I went to do a pseudo a PT update to update my repositories to see if I have any updates available for my raspberry pi.
CM four. It failed. It said, sorry, you can't update.
I was like, why? My time was off? Because I hadn't set anything up yet, but my time was not in sync and I wasn't allowed, wasn't permitted to sync with other servers, software applications, security.
You're going to have a hard time accessing websites if your time is not in sync with the rest of the world. And yes, you can get time hacked. Now, NTP is great for most things if all you need is millisecond accurate time.
But if you need more than that, I'm talking microsecond, nanosecond level accuracy. You got to get crazier with PTP or the Precision Time Protocol. This stuff is nuts.
But before we get into that, it's about time we took a coffee break and showed some love to our sponsor nor DPN. Now I use nor DPNA lot when I travel. For example, last year my family and I went to Japan in November for an entire month.
And while I was there I wanted to watch some US based shows. But some things are geo locked. Sorry, you can't watch the show from Japan.
You have to be in America. And I said, fine spun up nor VPN and I connected to an American VPN server. So suddenly I'm using that IP address, but while I'm sitting there in Kyoto eating ramen, it thinks I'm an American.
I can watch whatever I want. And now that I'm back here in America, there are some things in Japan that I also want to watch Japanese tv. I also use the word VPN to protect myself from myself because often when I'm doing things in my videos, I will be showing IP addresses, specifically my public one and it has leaked before.
So normally I try to use the word VPN to hide my public IP address when I'm doing things in my videos. When you're visiting websites, it'll see you as a server owned by nor VPN and not you at your house on Miller Drve. And while a lot of websites are secure with HTPS and TLS certificates, not everything is, how do you know?
I mean, you just don't. But if you use Nor VPN, they'll encrypt all your traffic so that hackers will have a very hard time seeing what you're doing. I almost shared my IP address trying to show you this page.
See, we're having to blur this out right now in my location. So check out my link below nor vpn. com/network check and you get a pretty good deal, okay?
$3 85 cents a month and you get all these features and it's time for this coffee break to be over. It's time for us to get back to talking about time. Now, before we do anything, I want to put this inside my server first.
I don't like it sitting here. I feel like it's going to break. Let's do that right now.
So I'm putting my time server inside my server room, specifically inside one of my prox mock servers provided by 45 drives. They're amazing. So I'll pop the lid and escalate safely inside one of its PCIE slots.
Now I don't have to power off my server or anything. All this thing is doing is pulling power from PCIE. So I'll tuck him in, screw in my antenna.
And now let's see how he's doing. We'll get logged in and the first thing we'll check is the GPS signal. I'm actually kind of worried about that because I have my antenna in my server room.
I'm not sure how it's going to be able to see this guy and talk to satellites. And here's how I find that out. I'll use a tool called minicom.
This allows us to view the output terminal output from our dev TTYS zero interface. It's a serial interface. Now why that without GPS module?
The G-P-S-G-N-S-S from U blocks, it's receiving its data from satellites and then feeding that into some pens on the Raspberry Pi that's outputting to, I think it's pen zero and it's outputting to dev TTYS zero. So I can read the raw GPS data coming in. Let's go take a look.
Jump in the minicom. minicom is kind of the worst. Do control A ZP reports.
I'll do C for 9,600 and go. Okay, so right now you can see the live GPS data coming in and it's not good. It's what I kind of expected.
You see all those nines, that means I'm not really getting any signal. I'm not locking onto any satellites. I was hoping I wouldn't have to do this because the antenna is just sitting on top of my rack.
It's magnetic kind of cool. So I either have to buy a really long antenna cord and put that sucker outside on the roof or something, which I don't want to do. I could try moving it out of the closet and put it in my main area, my coffee shop area and see if I get a better signal there.
But I'm going to try it here first because I don't want to have to go out there. Alright, I'm going to get my ethernet plugged in. Slide this sucker into the zema board and wait for it to come up.
Now I'm not going to plug the antenna just yet. I want to see the GPS happen as I plug it in. Be kind of fun.
We'll get back in a minicom, go through this crazy menu. Same terrible GPS settings. Now let's plug in our antenna sitting over here on the couch.
Great signal, I bet. Alright, let's see if anything happens. You do have to give it a few minutes and I did do that for the server room to make sure it gets a lock.
Goodness, nothing yet. Maybe it has to be outside, Man, nothing. Okay, tell the troubleshoot I'm going to boot it up with the antenna plugged in.
Maybe that has something to do with it. I don't know. Just pop it off and pop it back on.
I'm going to try reseeding it because I did just film B roll and we'll see what happens. And what I mean is when I was filming B roll, I took it all apart. So maybe it just needs to be reseed common troubleshooting tactic and hopefully I didn't break anything in the process.
Still nothing. Alright, I want to put the antenna somewhere higher on my filing cabinet. Look at my time right now.
It's 1970. Oh man, I need a signal stat. Okay, so I put the dang thing outside.
Let's see if we got any data. Yes. Now I don't know how much this data I can show you because is that giving my location away?
I don't know. I'll ask Chad g Petit later. But that's what it looks like when I'm getting all sorts of data from all kinds of GPS satellites.
Perfect. So we know that the GPS module is getting data to our raspberry pie. Now what do we do with that data?
And that's where the time beat software comes in. They not only make the device, but they have a pretty awesome PTP software. Now this sucker came pre-installed and most of the config is managed in a YAML file time beat yam.
I won't get two in the weeds here, but I'll show you what I'm doing and it's kind of key to how PTP works. Well some of it's now, by the way, this time beat software, it can be used for both server and client. So while my raspberry pi here is the server, I'll be installing time B on my other servers to receive the time from it.
There are alternatives like PTP four Linux, but I think time beat is better at getting you that nanosecond level accuracy. Actually, let's talk about that for a second. Nanosecond level accuracy, what does that mean?
And it takes a minute to kind of wrap your head around how crazy that is. So Ian explained it to me this way to help us understand the massive gap. So we have one second, which is a thousand milliseconds or a million microseconds or a billion nanoseconds.
And Ian said it like this. If you assume the world was built in seconds, 5 million seconds, it lasts for 12 days, I'm pretty sure. Yes, 11.
57 days. So let's round that up. So a million seconds is 12 days, a billion seconds is 32 years.
And that makes it easier to understand that gap. You've got Elon around like a million and a billion gets thrown around a lot, right? All over the place.
People can't just link is oh, I've got a million, I'll get a billion next. It's just the next step. But when you think of it in that length, like 12 days or 32 years, you wouldn't even go that far, right?
Like 12 years maybe you might think, right? Or a year. But yeah, 32 years.
That is crazy would be. So that's the easiest way I can explain the 32. Years.
So most people with time and NTP are getting accuracy within milliseconds. With my time server. We're jumping right over to nanosecond level accuracy.
Lemme show you how it's happening. So here in my time beat software, I have my primary clocks, the clocks that I'm defining for my Raspberry Pi server to receive time. The primary protocol is PPS.
Now what is that? This is pulse per second. This is a signal or a pulse being sent from the GPS module to my Raspberry Pi once per second.
And it's precisely aligned with the information it's getting from the GPS clocks. So think of PPS, kind of like the quartz crystal. In our computers we have the quartz crystal.
It's vibrating 32,768 times per minute or second, I'm sorry, back in the day we had pendulums that would go back and forth. We and our systems keep track of time by those regular intervals and so does our raspberry pi. Now let's pulse per second.
It's receiving it from GPS and these GPS satellites with atomic clocks on them. We can consider that UTC time or coordinated universal time. I don't know who did that.
They really need some help with acronyms. So when you're asking what time it is, you're really asking what is UTC right now? What do our atomic clocks the sources actually say.
So our PPS is drinking straight from the source UTC atomic clocks and down here cylinder primary clocks, we have our PTP set here. We could configure it to be a PTP client where it would receive time from a PTP server. But right now this is our guy.
He's a server only. We also refer to him as the grand master, probably the coolest name for the main server in an environment. And here's the hierarchy.
My raspberry pie will be the grand master in my PTP environment. What that means is he has the most accurate time source out of any server or device in my network. And PTP will actually use mechanisms to figure that out.
It's largely multicast based. So essentially a bunch of messages being sent across the network going, Hey, hey guys, who has the best time? It's like you in a crowd asking everyone, Hey, what time is it?
And you check each person's time and say you're the most accurate. That's what's happening in our network when we're running PTP. You also might have other clocks in the network like a boundary clock.
This could be a switch and it's kind of like a middleman. It'd be receiving time from the grand master and then forwarding it along to clients. This is cool because it'll help reduce the load on your grand master if he's serving a lot of time to a lot of servers.
You also might have a transparent clock in your network too. This would also likely be a switch. And what it does is fascinating because it will actually look at the PTP packets being sent through it.
Measure the time it took to process those packets and then adjust the time on those packets before they're sent out to the clients. So it's super accurate. Now of course, a packet going through a switch that doesn't take a lot of time, right?
Not to us, but when we're talking about nanoseconds, that matters. Now there are other clock types and servers, but there's one cool thing about PTPI want you to know it's what makes it so fast. And it's all about this guy right here.
The ethernet port. PTP does a thing called hardware timestamping. And what this does is it'll capture the exact time a packet is sent or received and stamp it with the time on the Nick Hardware level.
This bypasses any delays that you might experience through software. And when you're talking about time, it's all about eliminating delays. Doing it in hardware, man, it's smoking fast.
Now sure you can do PTP timestamping and software not going to be as fast. So you'll want Nicks in your network that support hardware timestamping, especially for your PTP Grand Master. Now, the Raspberry Pi CM four, it does support hardware.
What's funny though is the Raspberry PI four does not support hardware, but the Raspberry PI five does. Just a little fun fact there for you. And the one more crazy way it makes sure all our stuff is in sync to level, the Grand Master will send out sync messages pretty often.
A lot more than NTP at the highest. You'll probably see every two seconds, but you can tune that to every second or go subsecond once every 125 milliseconds. The config is good.
And now we're going to start our time beat software. Here we go. Oh wait, I was doing status now here we go started.
Let's do a journal CTL to look at the logs and see how it's performing. Things seem to be going well. It is receiving GPS information and time.
Let's check our actual time, our system time, time date, CTL. Okay, now we're here in 2024 and not 1970. That's good.
Now I did ask Ian about this. This kind of freaked me out. It said system clock synchronized.
No what? It's okay. The time beat software, you have it configured to adjust our clock and it's doing that.
We don't have to worry about it. We saw it adjust our clock earlier when it adjusted us to 1970. Now I always say this, if you don't track your time, then you don't know what you're doing.
And Time BEAT provides a pretty killer way to track our stuff. They provide us with an elastic database, or I'm sorry, an elastic search database paired with a Grafana front end. Thankfully they gave me a Kubernetes helm chart to deploy it very easily.
And here it's so right here, here's my raspberry pie. I think it's still syncing up. Notice our offset is going down.
Now the offset is how accurate our time is compared to real time atomic clock time and it's going down. Our goal here is nanosecond. We're in millisecond range right now.
Okay, do you see this 0. 18 nanosecond accuracy. What even is that?
Now this is directly from our PHC or our precision hardware clock that's built into our nick and doing the hardware timestamping. So it's hardware based and it's pulling its time data directly from the source PPS that we can dive a bit deeper by jumping into our Raspberry Pi here and look at all our clocks. So here's our master clock right here actually, which is ethernet zero.
Then we have our system clock, which you'll notice if we look at the data here, it is slightly off because that's how much time can pass between syncing with the hardware clock and our system clock. Ah, I got logged up. Which again, not a big deal, but when you're dealing with really sensitive systems that require nanosecond level syncing, it can have a massive impact.
Here's our PPS and then we also have an NTP source as backup. This is from nist. Notice the sync, it's in microseconds.
So our grandmaster, he's ready. He's ready to serve up some time, not time to set up my servers. I'm going to choose one of my 45 drive servers here.
And first I'll make sure my nick can support PTP hardware timestamping, do IP address, see what my knicks are. I want to dedicate one to this. I'll do ENO one, make sure he gets an IP address.
Perfect. The tool will use a test that's called E eth tool. We'll do dash t specify our interface and it'll tell us if it can do hardware timestamping, which it can.
Perfect. And now time to install time beat. I'll grab the install file from Time beat, then we'll install it, do a little bit of configuration.
We'll change this disable to false, like a double negative here to make sure it does change my system clock when it receives PT P information. And then we'll change the interface we use to ENO one and un comment. I'll add one more piece of information.
I wanted to send its data to an elastic search database, the one we already have. Alright, I'll do a little test here using Time Beat. They'll test the config, make sure it's good.
It looks good. And then we'll start the application system. CTO.
Start Time beat. Let's check the status. Looking good so far I think, oh, you know what?
I ran into this issue when I was talking with Ian. I could not get my clients to sync as I was testing it. It was a firewall issue.
Yep, I was being dumb. So I need to allow PTP traffic in my firewall configuration. Three 19 and three 20 UDP.
We'll reload. I'll restart. Time beats.
No, not start. Restart. Check our status now.
Okay, it looks like we're receiving stuff right now. Yes we are. Awesome.
Alright, let's check the graph. Let's see if it pulls it in. Ah, there it is, Hufflepuff.
And it says zeros right now. It's not that good. It's still sinking.
Yep, there it goes. So right now we're at millisecond accuracy. We'll give it some time here.
Give it some time. Endless, endless jokes. So I did not have hardware timestamping enabled.
Just did that for my client. Oh my gosh, we're in nanosecond level now. Yes.
Oh, this is cool. Nanosecond level synchronization. I don't know why I'm so excited about this.
I don't really need it at all, but I want it. Okay, right now my stuff is looking really good. I got all my servers in there.
Everything's pretty much doing nanosecond level synchronization. That's so exciting. Now, I didn't mention and anywhere in this video, I don't think why people do this.
Now, I talked to Ian about this a lot because they have customers that buy their stuff that helps people sink to nanosecond and there's even picosecond stuff they're going to be doing. That's crazy. But who does this?
Well, many industries need this. So financial trading is like their biggest customer. You got to be able to buy and sell and trade at a fraction of a second.
I mean, literally nanoseconds. NASA uses their stuff. They're getting big and broadcast multiple sources of audio and video have to sync up.
Okay, I'm literally running out of time right now because my camera battery is dying. I don't want to change it again. So we're going to do this real quick.
One last thing I want to talk about, actually two things. First, meta was one of the biggest people who put this PTP timing thing forward. They open source to a thing called tap.
I think I mentioned it earlier if I did video editor take this up. They spearheaded the tap open source project Time beat helped a lot with that, but Meta has a ton of servers that have to stay in sync and think about it. Meta messengers that we would call it.
Now you have to have your messages in sync when you're being sent across. You don't want to receive a message from the future or from the past. You want to receive it right now.
And that kind of times seek enables it to happen. Now, last thing, man, my battery's about to die. I want to set up NTP for my own self, for my systems that don't support PTP and Camera die.
I'm only going to look at this camera right now. I'm going to use a software called croi, which is great for NTP. Let's see how it's doing.
Extremely accurate. It's pulling from my PHC or my physical hardware or my, sorry, my precision hardware clock. Now all that's left is to update my NTP source here in Windows.
Now I was thinking about opening this up to the public, but then I'm like, no, I'm going to get hacked because you can't hack tp. Alright, we'll just go to date and time, internet time, change my settings and we'll change it to Time dot Hogwarts Studio. I set this up on DNS in the back end.
You don't have to worry about it. Update now. Got it.
I'm now using my own Stratum one and TP server self-hosted in my studio. But my servers, the ones that are running Prox Box and my CEF file cluster, they need to be in sync and in time. And by the way, you can install time B on Windows and Linux as a client on many, many systems.
I might actually do it here in my Windows system. I want to use PTP. It sounds better than NTP, but I met my goal.
I now have a time server in my studio. And you know what? I might set up another one because they sent me a PTP mini grandma thing.
Lemme show you real quick. Oh, okay. Yeah.
They sent me this, which is essentially the same thing according to Ian. I'm like, Hey, what's the difference? You guys sent me two things.
This is also rocking a raspberry pi, CM four. It's actually on a raspberry pi compute module four IO board. One of the official ones, I think he said, the only difference is that this is in a box and you don't put it into A-P-C-I-E slot has full-sized HDMI.
It does require an external power source, but I'm going to run two of em. And again, it's using multicast my network. So as sending out these messages saying, Hey, we've got time.
We've got time. My clients will intelligently using PTP mechanisms go, Hey, okay, that one has accurate time. Oh, that one has.
And they'll choose the most accurate time. Anyways, it's time out in this video because it's taking too much time and it's time you hack that YouTube algorithm. Have you done that today?
Make sure you do hit that like button subscribe, notification bell comment. That's it, right? Yeah.
That's all. You got to hack YouTube today, ethically. Of course.
That's all I got. That's all the time I have. Ain't nobody got time for that.
Copyright © 2024. Made with ♥ in London by YTScribe.com