Just Because You Can Do It. Doesn't mean you should...

Lately I've been playing a great deal with Content Management Systems. Well one in particular: Ghost

I even wrote an article on how to host multiple instances of it on one DigitalOcean Droplet.

And it worked fine.

You can very easily run more than one instance of Ghost from a tiny virtual machine. The sites ran fast according to Google PageInsights. Never had any issues writing, publishing or accessing Ghost.

Until...

I went to upgrade the Ghost instances on darrenbeattie.com and dailytrainingsession.com recently.

Turns out I may have missed something important from the self-hosting requirements:

"A server with at least 1GB memory"

Now the $5 droplet on DigitalOcean has that, but I'm sure they mean per instance.

Turns out, while not a problem for running Ghost, it's a problem for updating Ghost. And trying to update using their very good CLI tool to upgrade either instance tells me there isn't enough memory on that sized droplet.

Good thing I hadn't gotten around to moving Skill Based Fitness to a new Ghost instance on the same server. I may have discovered the limitations of the $5 droplet running.

I had thought about it, and desperately want to move it away from Wordpress. The Google Page Insights have been absolutely terrible on that Wordpress site for years and my pre-paid hosting will be running out soon.

My Bad.

It's no wonder, that no one had readily written about it and I had such difficult time piecing it together. Good learning experience all the same. Especially about NGINX and DigitalOcean deployment.

Ghost is a slick, beautiful writing experience, that runs blazing fast but is a bit of a memory hog it turns out.

I mean, should I really complain? It's only 1GB worth of RAM. The computer I'm writing this on has 16x that.

I probably saw it, I simply chose to ignore it because spending $5 a month for three sites versus $15 seemed like a steal. A cheap way to play anyway.

As it turns out, you may want to consider some workarounds for long-term maintenance. That is, if you ever want to upgrade Ghost.

Workarounds

There are a few things to consider...

Increase the Droplet Size

The most obvious one is to stop the virtual machine, and upgrade it to the next size (2GB) droplet at $10/mo. Then upgrading Ghost should work just fine, depending on how many instances you have on your droplet. No different really from running separate $5 droplets for each site. Save a more complicated config.

You could possibly even downgrade back to the $5 droplet upon completion but this will be annoying every time you want to upgrade ghost. They push new releases monthly. And you probably should upgrade to stable releases at least on a semi-regular basis. You're risking data corruption to an extent by messing around with this too much.

Good quick work around, but likely not ideal for me long-term. I want to simplify my approach, not increase the complexity.

Embrace the JAMstack Further

If/when you're using Ghost as a headless CMS (which DTS.com and DB.com are not currently) it basically becomes your back-end. Assuming you aren't building anything more complex than a web page/blog.

I suspect (but have not confirmed) I can deliver content to two different front ends via webhooks from one Ghost instance on a single droplet. You're basically just exposing the Content API to two different places. Use tags to split the content to its appropriate front-end location.

Then simply design two different front ends using Gatsby or Next (or both?) and statically generate the front-end using service like Netlify (who coined the term JAMstack) or Vercel (creators of Next.js). Both of which wouldn't cost me anything, assuming the traffic doesn't get crazy.

Unlikely for DB.com, but I'm secretly hoping it might happen to DTS.com. 🤞

If I don't burn out first, writing 500-1000 word workouts every day. I may need to slow the pace to keep the time investment reasonable or build something entirely from scratch anyway. I'm not ruling that out. In theory, it should work.

Currently the problem I'd run into trying to implement this is the heavy integration with Ghost's membership functionality for DTS.com.

There is no way currently to implement a membership functionality that is anywhere near as slick (as they've built) into a JAMstack implementation. 😞

Without building something from scratch anyway. It's in beta, mind you, so that functionality could disappear entirely or maybe version 4 will feature JAMstack member support.

There's no way to know.

I'd even consider Ghost.org hosting plan if I could have one well managed instance serving multiple front ends. I can dream right?

Prioritize

What do I need/want the Ghost writing experience most for?

It's probably not DB.com. It's a small low-traffic website to show off my developer oriented (occasionally business-oriented?) thoughts. I occasionally publish ideas here. I write on DTS.com daily and SBF.com (Skill Based Fitness) weekly.

It doesn't make sense to have a separate Ghost instance for DB.com, when I write so infrequently here. I simply wanted to retake my name domain that I had let sit dormant for too long. It should be the first thing served when someone Googles my name.

I can do that with a static front end and likely something else that isn't Ghost. I can then link to Medium or something for my writing or utilize a hosted but free JAMstack CMS like Contentful, DatoCMS or Netlify CMS.

All of which I've been investigating and have a terrible writing experience compared to Ghost. Contentful and DataCMS free tiers might be enough for a tiny amount of content at least but their pricing is absolutely ludicrous for a hobbiest.

They are also overkill. More like a full back-end API on demand, which could be fun to play with on some other projects.  

Right now I'm leaning towards Netlify CMS because it appears to be free without limitation – it ties directly into GitHub. While not the best writing experience, nor as easy to get started with, it still is a good developer experience.

At least it'd be a good way to add to my lack of commits lately.

🤷‍♂️ In any effect you'll probably see some changes to darrenbeattie.com in the coming days/weeks and eventually DTS integrated into SBF.com so it can be served on one droplet with one ghost instance.

Maybe one day I'll look into serving one ghost instance to multiple domains and as always I'll let you know how that goes.

*Note: This won't affect any remaining paying clients (we've spoken), you'll simply have access to the successor side project in a different capacity from most – Daily Training Session. I'm not abandoning coaching entirely, just moving it to the sidelines for now.

I can't begin to describe how emotional this is to anyone who has never started a business, let alone dissolved one.

For the better part of fifteen years my identity has been that of 'Coach.' Fifteen Years!

Thankfully it has nothing to do with the Coronavirus/Covid-19. I'm feeling for a lot of business owners at the moment and I'm really glad I decided to pivot more heavily into tech when I did.

The Issue of Scale

Flashback to 2011, I was working my dream job as the Head Strength and Conditioning Coach for a University Sports Team. Okay so maybe I was shooting for a pro team...

A bit later that year the University of British Columbia would significantly change their strength and conditioning department and there would no longer be individual strength and conditioning coaches for each team but one head coach to look after the 600+ athletes on every team at the entire school.

I was asked to put my name into the hat and probably had a legitimate shot at the position given my previous coaching performance. I loved coaching the team and I think I could have handled the responsibility but I also had imposter syndrome to the max. I was too young to deserve the position I already had, let alone this one.

I also felt like I couldn't keep up with the ~80 athletes I was managing already. I was doing the best I could with what I had, but 600 athletes? That's a lot...

Would it be worth dealing with the slow bureaucracy of a higher education institution?

Around the same time I met some people in the tech/start-up world and started devouring books on the concept of scale.

I was struggling to scale myself and these people knew how to scale.

I became preoccupied with the concept of scale. I was pretty good at what I was doing, but once the scale hit 60-80 people I was working 60 hours a week or more just to keep up. It wasn't sustainable. If I wanted to help more people than that, it was next to impossible to keep doing what I was doing.

Where to turn?

Koachable

Through blind luck or open enthusiasm (I could never tell which) I managed to convince an investor to give me a seed round of funding for a web developer to see what I could do about my scale issue.

The concept: let other coaches easily create online systems that allowed them to scale and train more people at one time.

If you spent some time upfront to build out some systems available online from anywhere, then scaling to 100, 200, maybe even more people should be easy.

See most 'fitness' software then (and even now) is just a prettier version of Microsoft Excel. The majority of coaches still default to it for exercise programming for that reason and I'm no different.

Once you've built your templates, no online app can match the speed at which I can craft a custom program in Excel. Excel isn't fast by any means, but the forms web apps use for exercise program creation are three to five times slower IME.

I actually had a broader desire for athlete/client management and more specifically things that weren't related to exercise at all, such as nutrition or sport psychology. Quick daily feedback loops were my focus.

It sounds counterintuitive to people who don't write training programs, but software for delivering exercises to athletes was (is) actually the least of my concern. That is the easy part of the job.

This was going to be a coaching platform for any kind of coach because why not? Coaches of all sorts have had such a big impact on me, shouldn't they get access? Doesn't it broaden my potential customer pool? Well...

Truthfully, I had NO IDEA what I was doing.

I don't think any first-time startup founder really does. You don't know what you've gotten yourself into, until you're in it. By that time you don't know which way is up and which was is down. You simply have to find a way to keep moving forward.

Especially while you're still coaching 25-30 hours a week and trying to launch a start-up in what little spare time you have. Ya that was a bad idea but as a practical person I needed something to fall back on and some income deposited in the bank.

Big thinking, big ideas, poor execution. I did almost everything I was telling my clients at the time not to do. I was trying to be everything to everyone and do it all at once.

It flopped. Hard.

Eventually I got that email from the investor, "we need to talk..."

I knew it was coming and it took the wind right out of my sail.

They were honest with me, I couldn't argue with anything being said and thankfully we remain friends. By that point I was more concerned about f🐬king those friendships up than I was about making this start-up a giant success. I hate disappointing others and my mind had shifted to loss-aversion.

If I was lucky it might appeal to tens of thousands of coaches. It was probably never going to be a billion-dollar business. I was scratching my own itch and as it turned out very few people had the same itch. Most are happy to maintain the status quo. And as I'd later figure out: most people are lazy. Sad but true.

"Pay you for the privilege of building out all my own content?"

No one explicitly said that to me, but I got this vibe from the majority of coaches on the platform. And you know what?? I don't blame them...

Building a three, four, or even a six-month coaching program is a much larger arduous process than building a work-at-your-own-pace e-course.

My grand personal vision was several different six-month programs – as a kind of choose-your-own-adventure thing. Needless to say, I never got there.

I look at my current exercise library of ~650 videos, and the dozens of nutrition/psychology modules I've written since Koachable shut down in 2015 and the amount of time I've spent crafting it all. It would make a lot of people cringe.

Even I've found it incredibly difficult to keep up a regular pace of creating and modifying content. It's a lot of thankless work. Yes after 4-5 years, I accomplished quite a bit, but it still feels woefully inadequate compared to my "grand vision."

Was I expecting these coaches to pay me $99/mo only to spend an entire year or two creating the content on their own? Where would they find the time to actually coach?

I think at its peak we had maybe 50 coaches using the platform. None of whom ever really used the system as I intended. Fewer still who were paying to use the software.

Meanwhile, I just kept adding features. I might need it to X someday! Reminder system? Check! Messaging system? Check! Want to add audio or video to something? Check!

If a coach in the system made a request, we built it. I had this developer on stand-by that I felt I needed to keep busy. It was a mistake.

The only QA was me. The only marketer was me. The project manager was me. Everything was me except development and the odd bit of design work we had done. I learned a lot about startups very quickly, but it was too late.

I had created a monster. A reminder app on steroids to nag people you were coaching.

As I know now, "wouldn't it be cool if it did..." is a terrible idea and saying no is the key to start-up success.

In the end, it was a giant sprawl of an application with a ton of features, a lack of focus, no apparent or sustainable business model and I was the only person truly using it.

Rather than let it die completely. I did what I probably should have done from the beginning.

Fitnack Evolved Out of Koachable

I'm anything but lazy. I don't always focus on the right things, or finish all the things I start and I tend to bite off more than I can chew but I'm not lazy.

My Hail Mary at this point was to use the existing Koachable software and focus my attention on the thing that I had been asking the other coaches to do:

Build and sell my own vision of an online coaching program.

If I could do that, it would prove the system worked. Then I could reopen Koachable to coaches with my pre-built content and eliminate or at least reduce the suspected choke point for scale.

We halted development, paused coach sign-ups and stopped supporting any coaches still using the Koachable system so I could spend a few months focusing on the Fitnack coaching program using what we had built so far.

And...it felt like it was working...

But ultimately it was too late for Koachable. Web applications need maintenance and after a few months without funding or a full-stack developer that just wasn't going to happen. I didn't have the technical skills to maintain it on my own.

I didn't give up, but I needed something easier to maintain as a less technical person. I could use a little HTML and a little CSS, so I moved the content I created to a new Wordpress (WP) installation. Bought a theme that looked good and hacked together a little membership system out of various plug-ins. It was nowhere near the functionality of Koachable but it was something.

This is probably what I should have done from the beginning.

At the time my thinking was, if I could build a system that worked and was making $$ then I could take those numbers back to an investor and reignite Koachable as a project. At the very least slowly build it up back into a small tech business I could run with a few contractors.

The numbers were good enough for me to fill hours I wasn't training, but never good enough to take back to an investor or hire anyone for anything other than minor contract work. I was still working 60+ hours a week.

Maybe this was an issue of only putting my toe into the water? Rather than jumping all the way in?

I kept running into maintenance and logistic issues despite WP. Plugins, themes even Wordpress itself would be upgraded and break stuff in the app or stop being supported. Stripe API's would change and subscriptions would need to be migrated. At one point my checkout just didn't work and no one could help me figure out why. It was down for weeks before a potential client brought it to my attention! You can't run a business like that...

And most of all that damn issue of scale!

I should have been spending 50% of my time marketing the app but I spent 75% of it building, editing and creating content and the other 25% in video calls with the clients using it. Spending hours each day editing and uploading videos was/is necessary but tedious and more than a little boring.

I fell back on what I knew. Designing custom fitness programs and giving them to people online.

Even after we started travelling and I dropped my in-person business altogether – another heartbreaking but "necessary" experience – I still couldn't get the scale I desired.

That mythical triple digit client number still feels just beyond of reach.

The Next Steps

You'd think I should be doubling down on Fitnack right now based on Covid-19, right? Well I didn't know we'd be dealing with this when I decided on the next chapter!

After 4 years of working on Fitnack it became clear to me that the original goal has eroded. At first it was a way to keep the Koachable legacy/dream alive. But now?

I needed to change, a new challenge and I wanted to work with a team again.

The original goal of Koachable was to help coaches scale, and even at Fitnack's peak I still wasn't helping 100 people. I wasn't having the impact I wanted to have.

I love coaching. I love helping people. I think fitness is my conduit for giving back. And I've come to love technology. Especially web technology.

After 9 years of toying with the latter, it was time to take it more seriously.

I enrolled at Lighthouse Labs in their 12 week web development course in Toronto.

Upon graduation, I seriously wish I could completely redo Koachable or Fitnack in Node/ Express/ React, in a reasonable time frame but I'm not that good...yet...

It's probably for the best. It's time for Koachable to die. The school was pretty upfront about this. They'd give me enough technical knowledge to get a job, but only about 20% of what I'd need to build a production level app on my own.

For that I'd need ~3-5 years of job experience and one month post bootcamp, I have a clear sense of why. It's a HUGE field that is constantly evolving and I'm nose down looking for the right opportunity with the right company as you read this.

At the same time, maintaining a wordpress site with these new skills seems silly. Pfft, I can tackle something far more complex than that, so I am.

I simply won't have the time to maintain Fitnack or coach as much as I have been, and work a full time development job. It's time to change my identity from "coach" to "web developer."

The New Digs

You can take a coach out of the gym, but you can't take that desire to help others out of the coach.

For my mid-term project at LHL we built a doodle clone. Nothing crazy but the first step of the design process stated: everything starts with DATA.

Not that Data...

• What is the data you can access or acquire?
• Can you correlate that data to something else?
• How can you pivot that data to add value?

Starting Fitnack meant I had accumulated this giant amount of data. I wasn't going to throw all that data away!

What could I use that data for that was more complicated than traditional Wordpress development, but less complicated than a full-stack web app with hundreds of features?

One that could be maintained on evenings and weekends, but would keep the data train slowly rolling? One that could potentially reach thousands of people and keep me a little connected to my roots in strength and conditioning?

I give you: dailytrainingsession.com

The Idea

The premise is simple.

Sign up for the list. And every day at 4:30AM Eastern Time (starting sometime soon, I'm shooting for May the 4th...be with you) you'll get an email with a workout I designed for that day.

It's not a WOD, it's a DTS. And it's free.

Think a smarter Crossfit WOD, featuring an actual plan (i.e. not random) for each month. I've got a pretty big collection of videos I can share so you can see the technique.

And I can add some quick commentary (my goal is to keep it < 1000 words, probably much less), give you equipment recommendations, and share some research/articles by other professionals I think you should read.

For the time being I'm going to keep it focused on short at-home workouts (for obvious reasons) with minimal equipment. In the future? I want to give you an exposure to a variety of training methods and approaches I was previously perfecting.

If you want to throw a little shade my way to help pay for the servers that's awesome!

You'll also get more input into the programs I design, and some exclusives so you can ask me questions, get modification suggestions and access the back catalog. That's going to be measly $5/mo or $50 a year.

*Don't sign up for the paid plan yet though, just put your name on the list and wait. I'll let you know when I formally launch via email.

I'll keep using my new dev skills to push the exclusives in a direction the community prefers on my evenings/weekends and I can hone those skills further with a day gig at an awesome tech company in Vancouver.

Deal?

Photo by Timothy Dykes on Unsplash

For a three month-long code bootcamp I've had the following (taken from a written lesson if I recall) scribbled on the front page of my notepad.

It's a step-by-step thought process to use when you're stuck coding (with some personal modifications).

HAVE I???...

• ...really understood the problem??
• ...pseudo-coded a solution?
• ...googled the error message/ related or similar problems?
• ...actually READ the error message or the documentation?
• ...double-checked my syntax?
• ...linted my code? (rarely in my experience so far has this been the issue...)
• ...rubber ducked the problem?
• ...pair-programmed or had a peer review my code?
• ...walked away from my computer and taken a break??

At any point I'd say you could walk away from the computer and take a 10-30 minute break. I've had more epiphanies away from my computer so far, than I have had trying to brute force a solution by gluing my butt to the chair.

If you don't know what rubber ducking a problem is – I certainly didn't before school – you basically talk through your problem with a rubber duck. It's a thing in comp sci. Basically it forces you to verbalize the issue and talk it through. Talking to an imaginary friend about the problem likely works just a well.

For the bolded pair-programming/ code-review I'd argue it's a lot like coaching and bringing in an external pair of eyes to look for things that you cannot see.

Human beings lack really effective personal awareness systems. We fail to see a lot of what we do. That external pair of eyes is paramount, so in the absence of mentors (from school) I've been using CodeMentor. I don't get anything to tell you that, I just think it's cool.

Sometimes (often times?) you just have to move on and come back to something with fresh eyes.

I used to wonder why so many tech companies had video games, ping pong or pool tables in the office. I don't wonder anymore. When you're thinking that intensely, staring at a screen for too long the most effective thing you can do is stimulate your brain in an entirely different way. I don't favour video games in this context myself but physical activity I completely understand.

Seriously, you don't know until you've tried it. Regular breaks are useful, despite my reluctance to take them.

Mostly I just wanted to tear this page off my notepad and put it somewhere else for reference. This seemed like a decent place. 😉

Photo by Meritt Thomas / Unsplash

Alright so I set everything up in my last post and everything was working just fine until I went to set up ghost memberships on dailytrainingsession.com and the signup process wouldn't work.

Ghost is using a cool passwordless approach to authentication. That means no passwords for anyone to remember! This eliminates the potential for any password data theft and thus limits the liability of site owners.

It also means that you login with your email only, then you have ten minutes to check your email, and click that email to get the appropriate JSON web tokens (JWT) for authentication.

It's a cool work around for passwords but it turns out it relies strongly on the Stripe and Mailgun integrations to effectively serve the JWT.

I clearly didn't configure my base URL properly and Stripe was refusing the request over http, so I had to find a way to force https.

I was googling and googling and going down a rabbit hole of complexity and things I didn't really want to have to do. Until I luckily stumbled across this documentation that was in front of my eyes the whole time.

It turned out to be a lot simpler than I initially thought it would be. Now I have to double check my Mailgun configuration because the email isn't getting through but that's a new post in the near future.

Here's what I did (you may want to review my last post if it doesn't make sense).

Step 1: Login to Your Droplet/Server

For beginners the '$' in these can be omitted, it's just a common way to write out the input of the terminal.

$ ssh root@your_server_ip

Step 2: Login to the Appropriate User

In that last post my first user was called sammy, and my second user for my second site was named janey but you should replace either with the username that you used to set up the domain you need to change.

If I wanted to change the configuration of my first site it'd look like this:

$ su - sammy

You should see your command line prompt change to something like:

sammy@ubuntu-nameofserver:~$

Step 3: Find the Appropriate Domain Configuration

If you know it (I have mine written down) you can simply navigate right into that directory:

$ cd /var/www/[nameofdomain]

If you forget or simply want to see what domains you have configurations for:

$ cd /var/www

To list the directories on your screen it's a simple command that everyone should know:

$ ls

Step 4: Run the Ghost-CLI Commands

From here on down it's basically just following that ghost documentation:

$ ghost config url https://[yourdomain].com

Then if you didn't set up your ssl in that last article for some reason you'd run:

$ ghost setup nginx ssl

This is to set up the ssl but we already did, so running that command is redundant and won't do anything else.

Step 5: Restart Ghost

Simply run:

$ ghost restart

You'll probably have to enter your user password and you're not done yet but you're close.

Step 6: Set Up a www. SSL Nginx Config

This is a little buried based on our previous set up and you won't be able to just adjust your global Nginx configuration.

The way we configured multiple ghost domains on one digital ocean droplet is that we use the ghost-CLI tool to generate a Nginx config file for each domain.

Note: You'll still need to be logged into the appropriate user you set up for any individual domain you want to force https to. Meaning you'll have to do this for each individual domain under each different user you've associated with that domain.

My Problem

The problem that I ran into after doing the above was that if I typed in or googled http://www.darrenbeattie.com specifically and it would give you a warning that this was insecure.

This was despite setting up my www A name properly on Digital Ocean and seems be a minor flaw in how the ghost-CLI tool sets SSL up on your behalf.

Type in www.darrenbeattie.com and it was redirecting properly to https://darrenbeattie.com. Type in http://darrenbeattie.com and it would also redirect properly, it was just that one specific path.

Happy Path: How likely is it these days that a user will explicitly type in http://www.darrenbeattie.com??

Probably not that likely, but if it was showing up strangely on Google I still didn't want to risk it. Especially on any of my other sites where https is even more important because they are taking money via Stripe.

The next step was to follow along to this section of the ghost documentation.

"You may wish to have multiple domains that redirect to your site, e.g. to have an extra TLD or to support www. domains. Ghost itself can only ever have one domain pointed at it. This is intentional for SEO purposes, however you can always redirect extra domains to your Ghost install using nginx. If you want to redirect an HTTPS domain, you must have a certificate for it.

Basically we have to set up a second SSL to www.[yourdomain].com and we do that like this from your /var/www/[nameofdomain] file so that you have access to ghost commands:

$ ghost config url https://www.[yourdomain].com

Don't worry, this will not restart Ghost or alter your https://[yourdomain].com configuration. You're merely switching over to a new config from within your domain file. You should see something like:

Successfully set 'url' to 'https://www.[yourdomain].com'

And I believe (though I'm not 100% sure) that because each user only has permissions for each domain that this is partly why the following commands will all end up where they are supposed to. 🤞

Next, get Ghost-CLI to generate another SSL setup for you:

ghost setup nginx ssl

You’ve now got two domains setup with SSL. The main canonical domain, and your www. domain.

Now change back to your canonical (main) Ghost config back before you forget.

ghost config url https://[yourdomain].com

Step 7: Change Your Ghost Nginx Config Files

Finally, you’ll need to edit the Nginx config files for your main canonical and your new www. domains to redirect properly.

*Note: The first time I published this article, I had only changed the [yourdomain].com.conf, which explains why the other redirects worked as they should.

Edit both /system/files/www.[yourdomain].com.conf and /system/files/my-[yourdomain].conf to redirect properly:

$ nano /system/files/www.[yourdomain].com.conf

Right under the server_name in the server object add the following:

server {    
    
    return 301 https://[yourdomain].com$request_uri;
}

Repeat with the other config file:

$ nano /system/files/[yourdomain].com.conf

And make the same change under the server_name in the server object:

server {    
    
    return 301 https://[yourdomain].com$request_uri;
}

Once you have made those changes run sudo nginx -t to get nginx to verify your config, and sudo nginx -s reload to reload nginx and pick up your new configuration.

Step 7: Finalize All Your Changes

Now use Nginx to verify your new configuration

$ sudo nginx -t

And reload your Nginx with the new configuration

$ sudo nginx -s reload

And you may as well restart ghost again just in case:

$ ghost restart

Wrapping Up

Well that turned out to be a lot easier in the end than I anticipated. That's always a nice surprise. dailytrainingsession.com signup is no longer spinning and giving me an error in the developer console. And hopefully I've resolved any and all straight redirects from unsecured http.

I get the confirmation that an email was sent:

My new problem is that the email is nowhere to be found!

Time to debug Mailgun...

Update: Fixing the https and restarting everything fixed the Mailgun issue. I suspect serving it over http initially meant Stripe wouldn't authenticate which meant it was never getting to Mailgun.

Photo by chuttersnap on Unsplash

You can absolutely do this tutorial still but you might want to choose a bigger Droplet than the $5 one listed below. I ran into the issue when attempting to upgrade Ghost. I just thought you should know before you get too deep into this tutorial.

Don’t Run Multiple Ghost Instances on a Small Virtual Machine

I found out the “hard way” why most people only host one Ghost instance on a DigitalOcean droplet. This is why they are called experiments.

Darren BeattieDarren Beattie

So the first time I set up darrenbeattie.com on DigitalOcean I made some mistakes. It worked but using the DigitalOcean one-click set up for the initial blog made it difficult to add additional domains.

I spent several days trying to make it work and I finally found an article that sort of detailed what I wanted to do. Sadly, the easiest solution was to blow up an existing droplet and create a new one.

The nice thing about $5 droplets is that they give you an inexpensive way to play. Sadly during my web development bootcamp I did not learn a ton about servers and deployment. I can code, but getting it up on a Ubuntu/Nginx server all by myself?

That's new...

I'm going to condense and update that article for you in my own words to give you a little more detail/specifics.

What To Do

I'm going to assume you have purchased your domain(s).

The majority of registrars only have good rates for a year, then the fees climb a ton. I had one domain climb to $50+ a year to keep. Others nickel and dime you until you end up paying triple or quadruple their initial offer.

I recently moved all of my domain names over to Google Domains (GD) for this reason.

As a Canadian it's a $17 CAD flat rate with private registration and it's very easy to use without all the annoying upsells of nearly every other provider (GoDaddy, HostGator, BlueHost, etc...) I've used in the past.

It's a good sign if I'm telling you that and I don't get any referral $$ for sharing that info with you. That's how annoying I've found nearly every other registrar prior to GD.

Step #1 - Sign Up For DigitalOcean

You should have some basic understanding as to how the terminal works on either a mac, linux or a windows machine.

You can sign up here. And if you do, this time I will get a referral bonus of $25 but you'll also get $100 worth of credit for 60 days or two free months if you choose the $5 droplet and I don't see why you wouldn't.

Unless you already have a gigantic audience the $5 droplet will easily handle most of the traffic of a small blog. I spent the extra $1 a month for weekly backups but you don't have to if you don't want to.

At the time of writing the process looked like this:

1. Pick the Ubuntu 18.04 option for your distribution
2. Pick 'standard'
3. Scroll left for the $5 droplet.
4. Choose your location (I chose Toronto because as I said, I'm Canadian)
5. I didn't choose any additional options but that's up to you
6. I set up SSH authentication (mentioned below, DO will walk you through it)
7. You probably only need the 1 droplet
8. Choose a hostname
9. Add any tags you think are relevant
10. Select the project (you'll need to create one if you haven't already)
11. I added backups, you don't have to, but it's $1 a month on a $5 droplet
12. Double check your selections and click create droplet

Yes, I get a kick-back this time, but similar to GD, DigitalOcean (DO) has an upfront no-BS approach to business. It's a simple service without the incessant upsells. We used them when we were running Koachable years ago and they helped us easily scale when we needed to.

You can use AWS or Google Cloud or Azure if you want, they are often free for the first year but then they can get expensive quickly past that. Those services tend to serve medium to larger sized businesses well but not tiny startups. We actually switched from AWS in 2014 to DO for this reason. I'm sure DO is more expensive when your traffic is huge but for now mine isn't.

I think it's wise to set up SSH for your droplet, but that's just me. It's more secure and it's annoying to type in your password every time you want to access your droplet (AKA virtual private server). For more on that...

How to Set Up SSH Keys on Ubuntu 18.04 | DigitalOcean

SSH-key-based authentication provides a more secure alternative to password-based authentication. In this tutorial we’ll learn how to set up SSH key-based authentication on an Ubuntu 18.04 installation.

Hanif JethaDigitalOcean

Although you may still want to use a passphrase from your favourite password manager, so you'll still be entering a password when you first login to your server each time.

Step #2 - Point Your DigitalOcean Nameservers To your Registrar

You'll have to point your DigitalOcean Nameservers to whatever provider you choose and you can read the following for more on that:

How To Point to DigitalOcean Nameservers From Common Domain Registrars | DigitalOcean

DNS is the naming system that is used to bind a domain name to a web server. In this guide, we will change the nameservers for a domain that we have registered and point it to a DigitalOcean Droplet.

Josh BarnettDigitalOcean

Basically you need to add these custom NS's under the DNS settings of the domain you want to use on your registrar's site:

ns1.digitalocean.com
ns2.digitalocean.com
ns3.digitalocean.com

Once you've done that, from within your project/droplet on DO you'll need to add the domain to DO and set up at least two A records within DO.

How to Add Domains | DigitalOcean Product Documentation

Add a domain to your DigitalOcean account to manage its DNS records from the control panel and API.

1. thenameofyourwebsite.com
2. www.thenameofyourwebsite.com

This ensures you'll be redirected appropriately no matter what people put into their web browser.

Step #3 - Initialize Your Server

Here is where I went wrong in my first attempt, at least for creating multiple blogs/sites on a single droplet. DigitalOcean has a 1-Click Ghost App Deploy set up in the marketplace and I just used that.

It works if you only want to create one site on your droplet. I wanted at least two, possibly more as I move some old wordpress sites over to ghost.

My mistake was not doing this myself and setting up users for each site. I used the root user instead, don't do that if you want multiple sites.

In fact, you shouldn't really do that even if you don't want multiple sites on one droplet. You should still set up a user with superuser permissions and use that instead.

The following is only partially correct in what we want but you should still read it. I didn't on my first attempt and regretted it.

Initial Server Setup with Ubuntu 18.04 | DigitalOcean

When you first create a new Ubuntu 18.04 server, there are a few configuration steps that you should take early on as part of the basic setup. This will increase the security and usability of your server and will give you a solid foundation for subsequent...

Justin EllingwoodDigitalOcean

Don't code along with that guide, I recommend coding along with me below. There are some differences.

Step A: Login to Your Droplet/Server

The '$' in these can be omitted it's just a common way to write out the input of the terminal. Experienced developers will know this, beginners might be confused.

$ ssh root@your_server_ip

Then you're going to get a couple of questions:

Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '<ip_address>' (ECDSA) to the list of known hosts.

You can just type 'y' rather than 'yes' if you want. If you are using SSH with a passphrase you'll be asked for it so make sure you have it handy.

Enter passphrase for key 'C:\Users\<you>/.ssh/id_rsa': <passphrase>

It's probably a best practice for security but it's up to you to implement it or not.

Once complete you'll be logged in to your droplet as what's called 'the root' and ready to get it set up. A mildly entertaining explanation:

Step B: Update The Server

$ apt-get update && apt-get upgrade

This may take a while and you'll get a ton of lines printing off.

Step C: Add a User

This example creates a new user called sammy, but you should replace it with a username that you want to use:

$ adduser sammy

I like to use acronyms for the domains I'm installing and I write the details down in my password manager.

You will be asked a few questions.

Enter a strong password and, optionally, fill in any of the additional information if you would like. This is not required and you can just hit ENTER in any field you wish to skip.

Step D: Make that User a Superuser

To unlock admin privileges for your new user we'll you'll need to do the following:

$ usermod -aG sudo sammy

Now, when logged in as your regular user, you can type sudo before commands to perform actions with superuser privileges. We'll need this, and we'll need to set up new users for additional sites per droplet in the same fashion.

Step E: Login as that User

$ su - sammy

Step F: Install Nginx

Nginx ('engine-X') is a reverse proxy. It's like the middleman between your Ubuntu (Linux) server and the rest of the internet. What makes the server do it's thing based on the requests really. A better back end engineer can probably correct me on that.

$ sudo apt-get install nginx 
[sudo] password for sammy: 

Enter your password and let it run. Then:

$ sudo ufw allow 'Nginx Full'

The second command allows HTTP and HTTPS connections. You should see a much faster confirmation on this command than the previous one. I don't know if that has relevance to any firewall but I've seen it mentioned a few times. In any case, this is what I did and it appears to have worked correctly.

Step G: Install MySQL

Ghost databases use MySQL so this is what we'll need. I'm far more used to postgreSQL and SQLite but they are all more similar than dissimilar so no big deal.

$ sudo apt-get install mysql-server

Once installed if you want to set a password you'll need to log into the MySQL shell which will look a little different in the terminal and we're switching to a structured query language (SQL) now:

mysql> ALTER USER 'root'@'localhost' IDENTIFIED WITH mysql_native_password BY 'password';

Replace 'password' with your stronger (stored-in-a-password-manager) password, but keep the quotations around your password or it won't work.

You should see the following after you change your password successfully:

Query OK, 0 rows affected (0.00 sec)

Then exit MySQL:

mysql> quit

Step F: Install Node.js

And npm...

Node.js is the server side JavaScript engine that will run your ghost site. As I said in my last post I really like node.

Anyhoo, we're installing the latest stable 10.x version using a curl command.

$ curl -sL https://deb.nodesource.com/setup_10.x | sudo -E bash -

Once that's done it will tell you something along the lines of:

## Run `sudo apt-get install -y nodejs` to install Node.js 10.x and npm
## You may also need development tools to build native addons:
     sudo apt-get install gcc g++ make
## To install the Yarn package manager, run:
     curl -sL https://dl.yarnpkg.com/debian/pubkey.gpg | sudo apt-key add -
     echo "deb https://dl.yarnpkg.com/debian/ stable main" | sudo tee /etc/apt/sources.list.d/yarn.list
     sudo apt-get update && sudo apt-get install yarn

You'll want to run the first command at a bare minimum:

$ sudo apt-get install -y nodejs

If you want to use the increasingly popular yarn package manager you'll also need to run the yarn related code too. But I'm happy with npm in this context and the rest of my instructions will use npm too.

Anecdotally yarn feels a bit faster, but I don't know how much npm or yarn I'll ever be using on this server so flip a coin ¯\_(ツ)_/¯

Step #4 - Set Up Ghost

The server is good to go, but you're not done yet if you want a ghost blog for your site.

We're going to use the Ghost-CLI command line tool to get Ghost installed and configured with the following npm package:

$ sudo npm install ghost-cli@latest -g

It's pretty quick. Next you'll want to make a directory (mkdir) for your site:

$ sudo mkdir -p /var/www/<yourblognamehere>

Replace <yourblognamehere> with whatever you want to call this file to keep track of your new ghost blog or website.

Then you want change the owner of this new directory to the user you created earlier:

$ sudo chown sammy:sammy /var/www/<yourblognamehere>

Again, replace sammy with whatever your username is and <yourblognamehere> with the name of your actual directory.

Set the correct permissions:

$ sudo chmod 775 /var/www/<yourblognamehere>

Then you need to navigate into that directory:

$ cd /var/www/<yourblognamehere>

This will change your prompt in the terminal to something like:

sammy@ubuntu-nameofserver:/var/www/<yourblognamehere>

That last bit helps you know that you're in the right directory and it's important that you're in the right directory. I always double check where I am before I run commands. It will save you time later.

From the directory (triple check it even):

$ ghost install

Now you'll get a bunch of command line prompts:

✔ Checking system Node.js version
✔ Checking logged in user
✔ Checking current folder permissions
✔ Checking operating system compatibility
✔ Checking for a MySQL installation
✔ Checking memory availability
✔ Checking for latest Ghost version
✔ Setting up install directory
✔ Downloading and installing Ghost v3.11.0
✔ Finishing install process

Followed by a bunch of questions:

• Blog URL → http://<your_domain_address>.com
• MySQL hostname → <default> hit enter
• MySQL username → root
• MySQL password → <your MySQL password>
• Ghost database name → <default> hit enter
• Do you wish to set up “ghost” mysql user? → yes
• Do you wish to set up Nginx? → yes
• Do you wish to set up SSL? → yes
• Enter your email (For SSL Certificate) → <email_address_you_want_to_use>
• Do you wish to set up Systemd? → yes
• Do you want to start Ghost? → yes

You'll get some check marks, with some other stuff along the way and hopefully everything goes smoothly for you.

I forgot to change my A server name stuff because I was moving an existing domain from one droplet to a new one. This led to an error on the SSL certificate portion, and it took me too long to figure out where I screwed up. Basically Lets Encrypt timed me out for too many requests and I had to wait an hour to fix that part before my site was live.

Lets Encrypt is a free SSL service that lets you run secure https:// connections. You'll need it if you want to use Stripe via Ghost memberships or any payment processor for that matter. But it's also just a generally good practice these days.

The Ghost-CLI documentation is your friend if you get stuck or have to troubleshoot any errors.

Now apparently if it goes smoothly it will prompt you to visit yourblognamehere.com/ghost but mine didn't. I previously knew I had to do this, so you simply navigate there in your web browser to get the ghost admin set up so you can start writing and playing with your site live.

Choose your theme make your changes and Bob's your uncle. If you only wanted one blog, adios amigo!

If not, keep reading...

Step #5 - Adding Additional Ghost Websites

Here is where things get a little trickier...

You'll need to logout from your current user situation (sammy in the example above) which is ctrl-d or you can simple type:

$ logout

Make sure you're using root so you should see something like:

root@ubuntu-nameofserver:~#

As your command line prompt. Or if you created a site and are just logging back into your server it should also take your right to root.

Step A - Create Another New User

To quickly reiterate:

$ adduser janey

Make that user a super user so that they have admin privileges:

$ usermod -aG sudo janey

Login as that new user:

$ su - janey

Step B - Set Up MySQL for New User

Open mysql this time with a slightly different command:

$ mysql -u root -p

When prompted submit your MySQL password from before (this is why you put these things in a safe place). We're going to use the same MySQL database here essentially but you need a separate user to keep your ghost websites separated.

Now that you're in the MySQL shell we need to create a new user. Again, replace janey with whatever you want to use* as your username and the password with whatever you want to use to secure it:

mysql> CREATE USER 'janey'@'localhost' IDENTIFIED BY 'password';

Remember don't remove the quotes, they are needed in SQL queries and you should get a 'Query OK' response.

*I simply used the same username for MySQL that I did for for my server user to keep things organized but I have a feeling this is not a best practice.

Next make sure you grant this new user the appropriate privileges and assuming you get a good response, you can exit MySQL with ctrl-d or type 'exit':

mysql> GRANT ALL PRIVILEGES ON *.* TO 'user2'@'localhost';
Query OK, 0 rows affected (0.00 sec)

mysql> exit

Step C - Set Up Another New Directory

Similar to before you'll need a new folder for your new ghost install.

$ sudo mkdir -p /var/www/<yoursecondwebsitenamehere>

Make sure change the owner of this new folder to your current user (second one):

$ sudo chown janey:janey /var/www/<yoursecondwebsitenamehere>

And grant yourself the right permissions:

$ sudo chmod 775 /var/www/<yoursecondwebsitenamehere>

Then navigate into that directory:

$ cd /var/www/<yoursecondwebsitenamehere>

Make sure you use your usernames and your site names rather than the ones supplied here.

Step D - Install a New Instance of Ghost

Once here you'll run the same command as before to install ghost but there will be several changes from the first time you did it because you're putting it on the same droplet as the previous one:

janey@ubuntu-nameofserver:/var/www/<secondblogname>$ ghost install

I'm just showing you how the command line should look in that example. The command is simply ghost install because we already npm installed it on the root.

This time after you get all your prompts answer the following questions like so:

• Blog URL → http://<your_second_domain_address>.com
• MySQL hostname → <default> hit enter
• MySQL username → <janey> (*or whatever your second username was)
• MySQL password → <janey's MySQL password>
• Ghost database name → <default> hit enter (*it creates a new db name for you)
• Do you wish to set up “ghost” mysql user? → No (*It should skip this...)
• Do you wish to set up Nginx? → yes
• Do you wish to set up SSL? → No
• Enter your email (For SSL Certificate) → <email_address_you_want_to_use>
• Do you wish to set up Systemd? → yes
• Do you want to start Ghost? → No

Notice the three No's this time? That's really important and you'll see why in a minute. We need to make some changes to this Nginx configuration file before we set up the ssl certificate and start ghost on this site.

Step E - Change Second Site's Nginx Configuration

Navigate into your Nginx configuration (config or conf) directory:

$ cd /etc/nginx/sites-enabled/

Then make sure you stop your Nginx server:

$ sudo service nginx stop

*Note: This command will make the server of your first site unavailable until your restart it again.

Open the Ubuntu nano terminal editor and modify your secondblog file:

$ nano <secondblogname>.com.conf

You can use the 'ls' command to list the conf files in this folder if you forget what you named the file exactly.

Change the <port> number (probably a 4 digit number currently) of the proxy_pass to an unoccupied but valid port number.  

You may have to troubleshoot this if you guess wrong, certain ports will be occupied but pretty much any 4 digit combo should in theory work as long as it wasn't the existing one.

...
location / {
        ...
        proxy_set_header Host $http_host;
        proxy_pass http://127.0.0.1:<port>;
}
...

*Remember the port number you choose you'll need it again in a minute.

You don't want two sites going to the same place if they are different sites, so you can't skip this step. If you configure this wrong, you'll likely be directed to site number one. Keep that in mind if you have to troubleshoot it after the fact. I'd start here with the port number.

When you hit ctrl-x to exit it will ask if you want to save. Make sure you say 'yes' and then hit enter to confirm the save before you exit.

Step E - Restart your Nginx Server and Set Up the SSL

Start your server again:

$ sudo service nginx start

Then navigate back to the second site directory:

$ cd /var/www/<secondblogname>

From here you'll have to open nano again on the config.production.json:

$ nano config.production.json

And make sure you change your config on the port to match whatever port you created on the second sites port:

...
{
  "url": "http://<secondblogname>.com",
  "server": {
    "port": <newportnumber>,
    "host": "127.0.0.1"
  },
...

Obviously put the port number you chose from before for this new set-up.

Now you can set the ssl with a special command from ghost-cli:

$ ghost setup ssl

Assuming everything goes smoothly you can now start ghost for your second site:

$ ghost start

Now everything should be good to go on yoursecondsitenamehere.com/ghost for you to walk through your admin set up and get that site up and running in a similar manner to the first. It's pretty self-explanatory.

If for some reason you run into a roadblock I encourage you to checkout the incredible ghost documentation.

Finishing Up

Phew...long article...

Alright that's it, I now have two ghost sites (working towards at least one more) up and running on one DigitalOcean droplet. Google analytics confirms that people can google them and other visitors have been this week.

Now I have to figure out how to force https:// without any errors...

Any suggestions?

Photo by Syarafina Yusof / Unsplash

I'm lucky enough to have owned my full name .com for more than 10 years now.

Many aren't so lucky, especially those with more commonly found names.

Not owning such a domain makes setting up a personal site when you're out looking for a new gig a bit more challenging. You're not quite as easy to look into and I'll come back to this in a moment.

At that time I set up a Wordpress Blog and I used it as a place to write about anything not related to fitness.

For those that care, I've been writing about fitness at skillbasedfitness.com since 2009. About 250+ articles now, many of them detailed enough I should turn them into mini-novellas and sell them on Amazon. I've also written more than 2500 answers/articles on Quora.com since 2012. Although I can't quite seem to crack 10k followers on that platform. My fitness-related work has been featured in Forbes, Thought Catalog, Medical Daily and Lifehacker.

A mistake switching hosting providers led to all of my content on the old darrenbeattie.com being lost. It wasn't much and a little thing called Medium had just come out (circa 2012).

Luckily, I had migrated some posts over to Medium. Rather than try to recover anything I lost on the migration I just redirected darrenbeattie.com to my medium site for years.

After all, I was a strength and conditioning coach, so having my biggest presence online be Skill Based Fitness made sense.

And in that time, some guy also named Darren Beattie became a speechwriter for Donald Trump and has been pushing his new book on Trumpist Nationalism under my very name!

Well, it's time to get this domain rolling again and push that *stuff back down the google search result page.

*There is now a 17-year-old hockey player in Canada with my name who thankfully helping me with this.

Now

I've spent the better part of 15 years as a strength and conditioning coach/personal trainer, and aspiring web entrepreneur.

I love it. And while most people switch careers because they hate what they're doing or there isn't enough money in it. Thankfully that wasn't me.

I've been a solo entrepreneur now for the better part of 9 years and things were getting a bit lonely. What was me was the need for a new challenge. I needed a way to liven things up again, to start working with a team again.

There was always one thing missing after I started Koachable and later Fitnack: Technical Skills.

What did I decide to do about it?

Well, I went to the Lighthouse Labs (LHL) and did their 12 Week Web Development Bootcamp. That's what.

Why did I do this?

Well to make myself immediately employable first and foremost. Without spending another 4 years in school working on a computer science degree. Impractical at this stage in my life.

Yes, I'm going to be winding down my fitness career, but hopefully, I'll have a little spare time to keep some of my existing online clients and maybe 5 hours a week, maybe on the weekends I'll do in-person coaching. You know, to keep that coaching muscle fresh.

It will become the equivalent of a side project (thankfully encouraged in the Web Development community) that I hope to tie many of my new found development skills into.

Think about it, this is not something strength and conditioning coaches do. Especially relatively successful ones. And that's what I like so much about doing it! It's so crazy it just might work.

It makes me fairly unique in both fields and development is something I've already been playing with for years on the side anyway. I hope that it helps me develop a very unique skill set in the long-term.

If you haven't read Range by David Epstein, read it and you'll understand my position better.

However, after 9 years of working for myself, my new challenge will be finding the right web development opportunity to take my skills to the next level.

LHL is pretty upfront about this. They give you enough skills to get a job, but not enough skills for you to come out of their program as a lead developer ready to build production-ready web applications.

That's another 5 years of learning on the job and thankfully I'm up to that task. Ideally a job with lots of growth opportunities and mentorship. An apprenticeship of sorts. I like building cool stuff that helps other people enhance their quality of life.

Which Brings Me to Ghost...

Here I was forwarding this darrenbeattie.com to medium willy nilly. Not getting any top-level traffic for my own name and it was time to bring it back.

Yet, I didn't want to bring it back to Wordpress (WP), which has gotten more and more cumbersome over the years. I have 3 other websites running WP currently and can confirm that the new Gutenberg editor on Wordpress is slow as molasses.

Honestly, it is a painful writing experience. It was designed to compete with Medium and you guessed it, Ghost!

I love the writing experience on Medium but they've long since decided to take their company in a different direction towards subscriptions and away from small bloggers like me.

Ghost's writing experience is similar but even better. And it's blazing fast!

Ghost is a headless CMS built on node.js. I just spent 12 hours a day, 6 days a week for 12 weeks in school learning what? Predominantly Javascript and you guessed it NODE.JS! In case it wasn't clear, I like node...

It occurred to me: What developer can't code their own blog? And at the same time, as I was repeatedly told in school, why reinvent the wheel?

Setting up a new ghost blog would permit me to test out some new skills on the back-end and the front-end. I'd have the freedom to code my blog in a manner I'm already a little familiar with, and at the same time, I can get a great writing experience without building one from scratch.

What'd I Do?

Ghost only charges you to use Ghost if you use their managed servers. At $29 a month, its not completely unreasonable but it's also not going to help me groove any back end skills either. It's also open-source and well documented.

Instead, I spun up a droplet on Digital Ocean for $5 as my back-end. If you click that link and sign up for Digital Ocean yourself, I get 5 months of hosting and you get 20 months if you also choose the $5 option. Good deal.

I configured my server, even managed to troubleshoot Digital Ocean's "1-Click" Ghost App creation (it's not exactly 1-click – i.e. I'm glad I learned some debugging skills in school) to get what you're reading right now. Version 0.9 if you will.

Ghost also lets me challenge my frontend code skills with handlebars.js templating if I want. Granted I haven't gotten there just yet (this is my first post) but I probably will. It looks interesting, but let's face it, not as interesting to me as React (my preferred front-end option currently).

More importantly, Ghost 3.0 brings a ton of support for the JAMstack. That means I can tightly integrate Ghost into a React app via Gatsby and/or Next.js. Two React platforms that I've been dying to play with.

It's all win-win. Over the next few weeks, I'm considering migrating my entire online eco-system to Ghost from Wordpress. And in the process, I'm taking control of all my website development in a way I never have before.

So far so good.