Forem: ADEKOLA Abdwahab

Sequelize, Postgresql column does not exist Fix

ADEKOLA Abdwahab — Mon, 13 Oct 2025 09:33:44 +0000

Fixing the “column does not exist” error in Sequelize + PostgreSQL

I woke to a slack message that an endpoint was hitting 500, an my check it was traced to Sequelize / PostgreSQL error, which I was conversant with while using PgAdmin but missed while using literal in Sequelize

This error:

column "isavailable" does not exist

If you are facing same issue, you’re not alone — and the fix is surprisingly simple.

💡 The cause

When you write raw SQL or Sequelize literals like this:

Sequelize.literal('CASE WHEN isAvailable THEN 1 ELSE 0 END')

Postgres interprets isAvailable as a lowercase identifier (isavailable) — which doesn’t exist.

✅ The fix

Wrap the column name in double quotes to make it case-sensitive:

Sequelize.literal('CASE WHEN "isAvailable" THEN 1 ELSE 0 END')

Now Postgres understands that you’re referring to the actual column name exactly as defined.

⚙️ Bonus Tip

To verify what columns really exist in your table, run this in pgAdmin or psql:

SELECT column_name 
FROM information_schema.columns 
WHERE table_name = 'property';

✨ Takeaway

Always remember:

PostgreSQL lowercases unquoted identifiers.
Wrapping column names in "" makes them exact and avoids this error.

Simple quotes, powerful fix. 💪

When Emails Stopped Sending After Migrating to DigitalOcean

ADEKOLA Abdwahab — Sun, 31 Aug 2025 11:10:36 +0000

A while ago, I had to move some services from Render to DigitalOcean Droplets. The migration went smoothly—until I noticed something odd.

Emails weren’t going out.

Every attempt ended with a timeout error. Frustrating.

At first, I suspected our email provider. A quick search even revealed a few people complaining about similar issues, so I confidently pointed fingers in that direction and proposed switching providers.

But then the unexpected happened—the exact same issue occurred with the new provider.
Meanwhile, everything worked perfectly on my local machine.

That’s when the lightbulb went on.
The issue wasn’t with the providers at all… it was with DigitalOcean itself.

The Real Culprit: Blocked SMTP Ports

After digging deeper, I discovered that DigitalOcean blocks all popular SMTP ports (25, 465, 587, etc.).
They do this intentionally to reduce spam, prevent abuse, and protect instance IP reputations.

Unfortunately for me, this meant any attempt to connect via the usual SMTP routes was doomed to fail.

The Workaround

Thankfully, my SMTP provider happened to support an alternate port that DigitalOcean didn’t block.
I switched to that port, and emails started flowing again. 🎉

But this was just a temporary fix.

The Better Approach

Moving forward, I’ve decided to avoid relying on SMTP altogether. Instead, I’ll be using email provider APIs for sending emails. They’re faster, more reliable, and not at the mercy of port restrictions.

Lesson Learned

What started out looking like a provider issue turned out to be a hosting environment limitation.
It reminded me of a simple but important lesson:

Always question your assumptions before blaming the wrong part of the stack.

And just like that, my brief urge to build a custom SMTP server quietly died. 😀

Circular Dependencies in NestJs

ADEKOLA Abdwahab — Sun, 15 Jun 2025 21:05:14 +0000

Circular dependency is a situation where two modules depend on one another.

One way to resolve this is to make just one of them depend on the other. So A ---> B, and B !----> A 😁

Another way is to use the forwardRef function provided in NestJs.

ForwarRef(B) inside A-Module

ForwardRef(A) inside B-Module

This will resolve most circular dependency issues; however, some are stubborn.

So, you need the silver bullet (🤦‍♂️):

Specifically Inject() the exceptional service(s)

'List' is not a subtype of type 'PigeonUserDetails?'

ADEKOLA Abdwahab — Mon, 23 Sep 2024 04:11:41 +0000

After I rebuilt my project to resolve an issue with a @freezed class my Google login feature started throwing errors.

'List<Object?>' is not a subtype of type 'PigeonUserDetails?'

I saw the issue on github, sadly the solution provided there did not work.

Doing some debugprints helped a bit but I could not see the status of the line that is failing after the last debugprint, hence I opted for the debugger, and then placed some breaking points.

I realized that there was no signInWithCredential method but there was currentUser which had the details of the logging-in user

//before
 UserCredential result = await auth.signInWithCredential(authCredential);

//after
 User? result =  auth.currentUser;//signInWithCredential(authCredential);

SMTP NODEMAILER TOO MANY CONNECTIONS

ADEKOLA Abdwahab — Tue, 06 Aug 2024 08:01:17 +0000

Straight to the point, if you encounter this error while sending emails via nodemailer module then you are likely sending too many mails while creating/using different connections simultaneously.

SOLUTION -

Configure your transporter to use the pool feature. See how below.

You should be fine by now, so let's go verbose.

When you make a request to a server, a connection is opened to that server. If you do not have other requests at that time, then you may forget abou the opened connection and go your way.

If you have another request, say instantly, you may also open another connection to that server. Meaning, you can open a different connection for each request that you want to make. Hence, from your authentication, you can have multiple connections opened at the same time - simultaneously.

It is possible you attempt that, and it is possible that the server limit how much you can open - the limit.

The question is, why can't we reuse a connection that we have already opened? To save resources. The art of reusing opened connections is POOLING.

Hence, instead of opening multiple, unnecessary, connections we can just open one, or a few, and reuse them as required.

Fortunately, nodemail supports this.

Vercel: Server Timed Out

ADEKOLA Abdwahab — Sun, 21 Jul 2024 15:01:14 +0000

Got this screenshot in my DM as the frontend developer on a project,some certain pages are not loading.

Upon investigating I realized that if the page was refreshed it would come up.

Essentially the affected pages are server-side rendered, and a recent push to the endpoints that serve them is taking so long to run.

Vercel has limits on running functions, depending on the plan that you're in, the invocation would time out and return the page above.

The backend needs refactoring, but we needed to mitigate this.

Vercel has configurations to change some defaults. You can increase the maxDuration timeout depending on the plan that you are

Created/update a vercel.json file accordingly, perhaps, as seen above.

We were fine.

How To Sort Array of Strings

ADEKOLA Abdwahab — Fri, 05 Jul 2024 19:07:49 +0000

I was building a pdf merger that takes two pdf files, merge them, then return a single file - result of the merging.

There is a requirement that the files must be in order, i.e. a particular file has to be at the top.

Since I exposed this via an endpoint, and I am using multer to manage file uploads as shown below

 upload.fields([{ name: 'pdf1', maxCount: 1 }, { name: 'pdf2', maxCount: 1 }])

Multer would not order the files, and there is no guarantee on how what I would get from Multer I had to resort to sorting the array of 'processed files' from multer. I sorted by 'filename.'

files.sort((a, b) => a.filename - b.filename);

To my greatest shock, the array was not sorted.

I hurriedly went to my terminal. launched REPL and tried it, then it got sorted on REPL.

😲

How? Why?

Then I went back to the basics of sorting, comparing, and returning -1, 0, or 1 if the first argument is lesser, equal, or greater than the second argument, respectively.

Like this -

Have you encountered a similar issue before? Or an inconsistency on REPL?

I would love to hear from you.

550 5.7.1 Relaying denied

ADEKOLA Abdwahab — Tue, 14 May 2024 05:54:11 +0000

Ooops

That's the error I got when I tried to send a test mail to a mailbox.

We observed that external emails were not coming in so we decided to test.

Message not delivered
Your message couldn't be delivered to [redacted] because the remote server is misconfigured. See technical details below for more information.
The response from the remote server was:
550 5.7.1 Relaying denied

The first thing that came to my mind was the DNS records, I went through them and confirmed that the MXs were good. I recently switched to the present DNS provider. So that is a suspect.

I researched the 550 5.7.1 Relaying denied and found out that it's truly related to DNS misconfiguration.

Went through the DNS records again, giving attention to the MXs, and realized that there is Mailgun mx record, we stopped/deactivated our Mailgun accounts a while ago. The suspect is pinned down.

I then proceeded to delete MX records that pointed to Mailgun, waited for some minutes then retried sending a mail test.

No more errors, mail well received.

WEB API VULNERABILITY THROUGH OTP

ADEKOLA Abdwahab — Wed, 08 May 2024 17:08:53 +0000

Broken Authentication is a popular part of OWASP TOP 10 list.

Authentication is one of the most important aspects of a platform.

It is the act of confirming that the user that wants to use the platform now through a particular account is actually the user that owns that account - assuming no user would share their login details.

The login details are registered into the system at sign up stage, most the details are password and email address.

Email address is regarded as WHAT YOU HAVE and password is WHAT YOU KNOW. The third factor of authentication is WHO YOU ARE e.g biometric.

When you put down your email address to sign-up, platforms need to confirm if truly you have that email. One way they confirm is by sending you a mail with a token/code which you are to copy and input into the provided place in the platform. If you are not able to do this then it would be considered that you don't have the email address, and your sign-up would not be complete.

Also after signing up it is possible for a user to forget their password, locking them out of their accounts.

Platforms provide means for users to get back their accounts by allowing users to reset their password. This is also achieved by confirming if the user that wants to reset the password actually HAVE (access) to the email box of the corresponding email address on the said account.

Typical content of such email that's sent looks like this

However going extra mile, hackers can sign-up or reset a password of an account for which they don't have the email address. This is possible especially when the platform is being accessed via web.

How?

Most of the codes that are sent are 4 digits. Okay, let's start from here: how many guess would it take to guess a single digit number? 10 guesses. How many would it take to guess two digits? 100. How many guesses would it take to guess four digits 10000.

Hackers, without having the email address, can just run from 0000 through 9999 and calling the API endpoint for each guess. They'd probably hit luck before reaching 9999. Hence defeating the check of WHAT YOU HAVE.

How many guesses would they have to make if the digits were 6? 1000000. Yeah that's one million guesses.

Hence the more the digits the more secure and impractical it becomes for hackers to guess.

With this the platform would be safer of broken authentication.

In one of my next articles I will write about whether to send a verification link is better than sending OTP/codes. What do you think?

How would you handle image upload?

ADEKOLA Abdwahab — Thu, 21 Mar 2024 12:48:42 +0000

User uploads profile picture.

Images are converted into URLs via third party asset storage.

Backend handles the conversion.

Which of the scenarios below would you adopt and why?

A. Frontend should call convertImageToUrl endpoint, then call updateprofilepicture endpoint with the image url.

B. Frontend should call updateprofilepicture endpoint with the raw image - so backend handles the conversion to url in that endpoint implementation.

What are your reasons?

How to Automatically Remove Console Logs and Other Code Snippets

ADEKOLA Abdwahab — Fri, 03 Nov 2023 02:26:00 +0000

Why Hidis?

Supports ES6 ✅

Supports JS ✅

Supports Typescript ❌ - coming soon

Supports CJS ✅

Hidis is a tool designed for developers who need to keep certain functions and code snippets on their local machines during development without committing or pushing them to remote branches (e.g., production).

Pre-Hidis Installation:

Hidis relies on pre commit hooks to achieve its goal.

Install Husky: Confirm that you have the husky package installed. If not, you can install it using the following npm command:

npm  i  husky  --save-dev

Setup Husky for Hidis: After installing Husky, run the following commands in your terminal, one after the other - take note of the differences for cjs and es6:

a. General to cjs and es6

npm  pkg  set  scripts.prepare="husky install"

npm  run  prepare

b. 🛑🛑🛑

If your project type is ES Module i.e you use import/export, run this 👇:


npx  husky  add  .husky/pre-commit  "node .hidis/precomm.js"

If your project type is CJS i.e. you use require(), run this 👇:

npx  husky  add  .husky/pre-commit  "node .hidis/precomm.c.js"

c. General to cjs and es6

git  add  .husky/pre-commit

Hidis Installation:

To install Hidis, run the following command:

npm  install  hidis  --save-dev

Hidis Setup

Hidis creates a folder named (.hidis) inside your root directory.

You need to import this folder only once but in your main /entry point file and in the first line. If not imported at first line then you will get Reference error, it is like you want to use hidis before its birth. 😆

ReferenceError: hidis is not defined

Import hidis like the below inside a file that you are sure would be loaded/required when the node server starts. It is called main entry point file. Your main file could be app.js or index.js. To confirm which file, check your package.json look for the value of main, that value/path leads to the file you are looking for.

Also ensure that the path you provides (import from) correctly points to the .hidis/index.js

*For projects in ES6 *:

import {} from "./.hidis/index.js";

*For projects in Cjs *:

let {} = require("./.hidis/index.c.js");

How does it work?

To use Hidis, start your Nodejs server, call the hidis function anywhere in your project. Hidis takes a string as an argument. Make sure to separate more than one line of code with semicolons. Here's a sample:

hidis(
  'console.log("Everything inside this hidis would not be committed"); let name = "suliyat"; console.log(`My name is not ${name}`)'
);

When you run git commit -m {your commit message} Hidis would hide hidis functions and their arguments from being committed, thus from being pushed.

Ghost Mode - Original Content

Sometimes you still need the hidis functions on your local machine after committing/pushing, so permanently deleting hidis functions at every commit might not present a good dev experience (DX).

To ensure good DX you have the option to choose if you want the hidis functions permanently removed or not. Run the following command to change the config. the default is true (i.e. returns hidis functions back into their places)

npm  run  hidisConfig  --  original [false |  true]

In ghost mode committed files would still appear 'unstaged', as if they have not been committed but they have been! It is just ghosting . . . 😁

Errors

Error: provide arguments: to resolve this error ensure the hidisConfig property in your package.json script is set appropriately - only one command should be there, just like in the sample below.

For this sample, the project is a cjs project - uses require().

"scripts": {

"test": "echo \"Error: no test specified\" && exit 1",

"start": "node index.js",

"dev": "nodemon index.js",

"prepare": "husky install",

"hidisConfig": "node .hidis/config.c.js"

},

Tweet at me https://twitter.com/wahabind

Always Return an Object - Why?

ADEKOLA Abdwahab — Sat, 03 Jun 2023 11:49:00 +0000

Look at this function:

function canShare(username){ .... Do some checks ... return true }

Code snippet 1

When we call canShare('uyut') we get true or false. There's no problem with this.

So you could call/use the function like this:

`if (!canShare) {
Do something else

alert('cannot share')
}`

code snippet 2

However when building API services, especially one consumed by third parties this kind of return could be a pain when the API grows.

With the snippet above the user only know they can't, there's no provision to let them know why they cannot or what action they need to take to resolve it.

To achieve this the function needs to return more than a Boolean but what...an object!

However with the initial type of return, couple of things would have to be changed to accommodate the change - and there'd be ripple effects.

We can edit the function to return an object of this structure:

function canShare(username){ ... return {success:Boolean, message: string, action: string} }

Code snippet 3

If an object is now returned the code snippet 2 would have to change to

`if (!canShare.success) {
Do something else

alert('cannot share')
}`

Code snippet 4

This looks like a simple change, however imagine the function being used in multiple ways, in multiple places.

What do you think?

Are there instances you think returning object would not fit?

What's your experience with issues like this?

Share, I want to read from you!