Me: GET /sleep
Baby: 307 Temporary Redirect
Baby: 204 No Content
Me: 200 OK

Me: GET /sleep
Baby: 307 Temporary Redirect
Baby: 413 Payload Too Large
Me: 102 Processing
Me: 200 OK

Me: GET /sleep
Baby: 307 Temporary Redirect
Baby: 444 Connection Closed Without Response
Me: 200 OK

Me: GET /sleep
Baby: 307 Temporary Redirect
Baby: 444 Connection Closed Without Response
Me: 429 Too Many Requests

Me: GET /sleep
Baby: 307 Temporary Redirect
Me: 101 Switching Protocols
Me: 408 Request Timeout
GF: 102 Processing

Me: GET /sleep
Sleep: 404 Not Found
Me: 406 Not Acceptable

(Morning)
Me: 501 Not Implemented

It fucking staggers me how many backend/devops-y people don't understand what a client side "request timeout" is, versus a server side one.

What does it mean:

The client was fed up with the servers bullshit, and decided to piss off and not wait around for the server to take forever to respond, because life's too short.

How not to solve/debug this issue:

- "I've checked the API request in tool xyz, and it works fine for me"

Congratulations, you've figured out how to call an API once, in isolation to the rest of the application, and without any excessive load. And using a different client to me, with a different configuration. Lets get back to actually looking at the issue shall we?

- "I only see HTTP 200's in the logs"

Yep, you probably will in most circumstances, because its the client complaining about it taking too long, not the server. If the server was telepathetic and knew what the client was thinking/doing at all times, we wouldn't have half of the errors we do.

- "Ah ok, I understand ... so how do I solve this?"

Your asking me? I don't fucking know, I didn't build the server! Put better logging in place and figure out why sometimes it takes forever.

Jesus fucking christ

Escalation channel

Q: @here I'm sending a request to X and I'm getting a timeout. Please help.

A: Can you please share the request you're sending?

Q: It's <URL> and I send this token as a Bearer: <token>

A: I understand that, but it's a lot easier to debug if we have the curl for said request. You can find it here: <postman_screenshot_with_arrows>

Q: Here's what it looks like: <postman_screenshot_with_17.4_km_long_curl>

I've tried multithreading with php, wrote a simple script which checks a series of ip addresses and tries to ftp into them.

I've noticed that the script is running very slow, i checked everything, tested the db, tested my code, i've started to doubt, that my compilation of php was fucked up (btw i did that for the first time).

Then i've started to mesure the time of each db request, but the numbers didn't add up. Then it fucking hit me...

I fucking set the timeout for ftp_connect to 5 seconds, and that was causing the slowdown. I wasted two fucking coding sessions on finding that out.

What a fucking blind moron can I be, holy shit.

The moment you realize that you have successfully beaten reality with your unit-tests...

There are unit-tests for ...
... the api returning a 408 Http StatusCode when an internal request times out.
... the react app take this status-code and fires an action to display a specific error message for the user.

Every bit of code runs just fine.

Deploy this hell of an app on the server. Dandy Doodle.

Do a smoketest of the new feature.
FAIL!
Chrome starts to crumble during runtime. The api Request freezes.
Firefox takes the 408 api response but fails to interpret it in react app.

So I began to wonder, what the hell is going on.

Actually I recognized that I had the glorious idea to return a clientside error code in a serverside api response.
Glorious stupidity :/

Finally I fixed the whole thingy by returning an 504 (Gateway timeout) instead of 408 (Clientside timeout)

Cheers!

I would like to call out the moron who decided to control docker through HTTP when the maximum time the server can take to finish the task is longer than the HTTP timeout.
If you expect things might time out, you don't use a HTTP request. You use a resource and poll it, or Websockets, or possibly SSE.
Shoehorning your API into a frame that obviously doesn't fit doesn't help anyone. Just admit that you don't know what HTTP is and use a regular TCP socket with regular pings.

FUCK APPLICATION LEVEL FIREWALLS!

So i cam online today, thought already lets open the shitty outlook webmail client. Holy crap .... thats way to much mails. Many of them are missed teams messages. So i open up teams and holy crap. Like every third dev in my company send me a message screaming "gitab is not working!!!".

Yesterday i updated it so imediately get in panic mode - what the shitty hack have i done?!

So yeah gitlab seems to be working just fine, everything is speedy and responsive, so i call one of my fellow devs and ask him whats wrong? And he is like oh yeah there comes a ldap error saying timeout or something.

I try to login with active directory. Works like a charm. Try another account, same problem?!

Google the problem, search gitlab tickets. Nope there is no open bug or sth. like this.

So alright lets call the network guy. "Yo, can you check if there is something ldap-like getting blocked to the gitlab server?" - He is like oh yeah damn like almost every damn request is getting blocked. Ah wait, there was an firewall update yesterday too. Yeah ldap is no longer ldap. BLOCK THAT SHIT!

After 10 minutes of figuring out what shitty type is detected by the firewall and what needs to be whitelisted to make it fucking work again it seems to work.

But ha no, there is another update rolling on, so same shit like 15 minutes later.

Now it seems to work and i have to inform every damn fcking developer that it works again. And yeah alright you sent a mail, but fuck it, i will call you though! So yeah just answering calls, mails and chat messages. Like why the fuck cant you read your mails like a damn normal person?!

Third day of working on my recruitment task, and I'm starting to get pissed. I'm applying for Junior JS developer (suprised that they even picked me, I had 1 JS project in my resume, rest was Java). The task seemed simple, create website with autocomplete field which gets 10 cities with most polluted air from given country and get cities deacription from Wikipedia. But hell no. First, the air quality API that they told me to use sucks horse dick. Like seriousy, you can get a fucking timeout while fetching data, because as author explained, someone decided to make 2 fucking queries per request, one to count all possible results, and then the second one for actual data. Like, WTF, why would you do that. After I got that shit to work from time to time, it was time to Wikipedia API. And the shitshow starts again. Because it turns out that you can't filter the results based on the category. Which means that if the city has the same name as river or some fucking guy doing sports, I won't get the fucking description, because it will simply return info, that there are more more that 1 result. At this point, I'm so fucking pissed, I am barely keeping it together. I want to work at this company, because the pay is great, there are a lot of opportunities and shot, but god dammit, if I finish this task, I'm getting drunk for 3 days straight.

EDIT: even author of the air quality API says that it is not a good fit for given task...

This had just happened, I was trying to increase the default timeout of an nginx running in a container for a proxy pass and always got a 504-gateway timeout response. I was setting proxy_connect_timeout, proxy_send_timeout, proxy_read_timeout, send_timeout, keepalive_timeout, etc. and nothing worked, after two hours of adding and removing lines of configuration (and waiting 1 minute for every time I tried a request), then I realized I have a local nginx for redirect server names to local ports (the container), that nginx was the one that actually responds with the 504 error, after that I tried a request with the port of the container ALL WORKED!!!!