I did a simple bar graph to show the status of a db migration. Each bar is a computer with a db file. As I was bored waiting, I added a King Kong hanging from the longest bar.

Running a huge migration on our production db.
Takes 45+15 minutes (migration+tests) on my lappy. So. I'm going to be bored and slightly worried for the next hour.

Hopefully the production servers will be faster.

On my former job we once bought a competing company that was failing.

Not for the code but for their customers.

But to make the transition easy we needed to understand their code and database to make a migration script.

And that was a real deep dive.

Their system was built on top of a home grown platform intended to let customers design their own business flows which meant it contained solutions for forms and workflow path design. But that never hit of so instead they used their own platform to design a new system for a more specific purpose.

This required some extra functionality and had it been for their customers to use that functionality would have been added to the platform.

But since they had given up on that they took an easy route and started adding direct references between the code and the configuration.

That is, in the configuration they added explicit class names and method names to be used as data store or for actions.

This was of cause never documented in any way.

And it also was a big contributing cause to their downfall as they hit a complexity they could not handle.

Even the slightest change required synchronizing between the config in the db and the compiled code, which meant you could not see mistakes in compilation but only by trying out every form and action that touched what you changed.

And without documentation or search tools that also meant that no one new could work the code, you had to know what used what to make any changes.

Luckily for us we mostly only needed to understand the storage in the database but even that took about a month to map out WITH the help of their developer ;)

It was not only the “inner platform” it was abusing and breaking the inner platform in more was I can count.

If you are going down the inner platform, at least make sure you go all the way and build it as if it was for the customers, then you at least keep it consistent and keep a clear border between platform and how it is used.

I HATE TESTING DB MIGRATIONS! SHIT TAKES BLOODY FOREVER!

This one takes 20 freaking minutes each attempt, and I need to run it. yet again.

$@%&!

I'm going on vacation next week, and all I need to do before then is finish up my three tickets. Two of them are done save a code review comment that amounts to combining two migrations -- 30 seconds of work. The other amounts to some research, then including some new images and passing it off to QA.

I finish the migrations, and run the fast migration script -- should take 10 minutes. I come back half an hour later, and it's sitting there, frozen. Whatever; I'll kill it and start it again. Failure: database doesn't exist. whatever, `mysql` `create database misery;` rerun. Frozen. FINE. I'll do the proper, longer script. Recreate the db, run the script.... STILL GODDAMN FREEZING.

WHATEVER.

Research time.
I switch branches, follow the code, and look for any reference to the images, asset directory, anything. There are none. I analyze the data we're sending to the third party (Apple); no references there either, yet they appear on-device. I scour the code for references for hours; none except for one ref in google-specific code. I grep every file in the entire codebase for any reference (another half hour) and find only that one ref. I give up. It works, somehow, and the how doesn't matter. I can just replace the images and all should be well. If it isn't, it will be super obvious during QA.

So... I'll just bug product for the new images, add them, and push. No need to run specs if all that's changed is some assets. I ask the lead product goon, and .... Slack shits the bed. The outage lasts for two hours and change.

Meanwhile, I'm still trying to run db migrations. shit keeps hanging.

Slack eventually comes back, and ... Mr. Product is long gone. fine, it's late, and I can't blame him for leaving for the night. I'll just do it tomorrow.

I make a drink. and another.
hard horchata is amazing. Sheelin white chocolate is amazing. Rum and Kahlua and milk is kind of amazing too. I'm on an alcoholic milk kick; sue me.

I randomly decide to switch branches and start the migration script again, because why not? I'm not doing anything else anyway. and while I'm at it, I randomly Slack again.

Hey, Product dude messaged me. He's totally confused as to what i want, and says "All I created was {exact thing i fucking asked for}". sfjaskfj. He asks for the current images so he can "noodle" on it and ofc realize that they're the same fucking things, and that all he needs to provide is the new "hero" banner. Just like I asked him for. whatever. I comply and send him the archive. he's offline for the night, and won't have the images "compiled" until tomorrow anyway. Back to drinking.

But before then, what about that migration I started? I check on it. it's fucking frozen. Because of course it fucking is.

I HAD FIFTEEN MINUTES OF FUCKING WORK TODAY, AND I WOULD BE DONE FOR NEARLY THREE FUCKING WEEKS.

UGH!

First company I worked for, overall it was a good experience, but at one point they promoted a consultant to project manager, and their planning skills were about as good as their people skills, which is to say, appalling.

We had a project update for a huge client, that required, for political BS reasons, that most of the team spend several weeks on-site, 300km away from home.
Go-live was approaching, and the plan was: migration starts Friday night, shortly after midnight (so actually, Saturday) once the client’s IT confirms DB is backed up. Expected duration: 5 hours.

- So, you expect me to work from midnight to 5am on Saturday? And when do we start working on Friday?

- 9am, of course.

- 9am!? So you actually planned a 20 hour work day? (Note: legal maximum here is 10 hours in a day, 40 a week)

- And we have to be there on Saturday 1pm to recheck everything is running smooth.

wtaf were they thinking?

...He hired a shit dev who did the same work in 3 times less than what I asked for.

He's now back crying to fix his Fuck up.

You ask how I know he is shit. He SSH-ed into the server. Worked directly off the production files. Worst of all, he installed phpmyadmin, changed the db structure without even writing a fucking migration !!!

How the hell am I supposed to know what he changed!! It's gonna be a long night 😥

OMG people please stop being so fucking lazy... help me help you... RN I have multiple support people asking me to fix a bug that they can't even describe (and honestly I doubt it exists) and a fellow developer who refuses to give me a DB or migration to test his patch but wants me to merge it urgently. FOH and die, y'all.

-Im a frontend
-We don't have any back-end in our team
-Im now a front-end & backend
We need to migrate our server to AWS but we have nobody
-im now a front-end & back-end & DevOps
During de migration we need to use AWS database and create new view and manage access
-im now a front-end & back-end & DevOps & DB engineer
-We have new employee (Yess)
-im now a front-end & back-end & DevOps & DB engineer & Trainer and repository manager (PR, Manager)

Public institution... No salairy growing... Fluck this shit

Contex: Working on a c++ frankenstein code (mixture of legacy and new stuff whith things depending on the client using it)

User Story: Migration from oracle to SQLite for half of the DB data

Summoner: One client wants to keep using legacy for now, therefore we need an strategy chooser templated singleton...

Satan 666 = Singletons + Static methods + Different compilation units

Result: 3/4 of the files of the full backend being modified for the migration.

Conclusion: When will be loaded on production company will probably lose many clients due to unspected bugs everywhere.

Insert potato here

One of our projects migrated their file-repository to another one during a major release.

Instead of giving this task to an experienced programmer, they gave it to the head of the respective dev department due to the usual release panic.

Soo.... He wrote the migration tool. It was executed during the release. Everything seemed fine so far.

A few days later. Someone from the above project came to my team due to some "strange behaviour on the production database".
They reported that they couldn't download some of the user's documents due to unknown reasons.

After quickly analyzing the current state of the new file-repository, we concluded that the affected documents did not exist in the new repository.

Then we took a look at the so called migration tool...
Well.. After nearly 30 min. we knew the root cause for that.
They only migrated the first 4 levels of the folder structure. Due to the assumption that "we don't use deeper nesting". (Facepalm)

As the head of their department wrote it, no one seems to questioned it either. Nor did they made a code review and ended up with a tool with hard coded urls to the production db, no version control, no build tool, no ci, nothing. Breaking nearly every possible company standard.

However.. That's not it. When analyzing their migration tool we noticed another even more dangerous thing.
They mixed up the id generation of the migrated documents resulting in a random assignment between customers and documents. Which is quite bad as this contains sensitive information. E.g. passports

They offered us quite a nice amount of money to fix this until EOB. We declinded as it was simply not possible in that time, but agreed to support them with the new tool.

After some time I heard that they migrated production again. And they fucked it up again. They never talked to us after we offered them support...

The third and final migration was written by us. Not only migrated it correctly. It was also way faster. By factor 20.

In the end we haven't gained anything from this rushed project as the penalties were piling up due to this fucked up migration.
After all this time I'm not sure who is to blame. In my opinion, partly all of them.
Head of department who can't and shouldn't code.
Seniors who didn't review the code and didn't ask for help.
Release mgmt who put way too much pressure on the devs.

One of the largest companies on the continent. Uses Oracle on AWS RDS with the beefiest resources available. It comes to the point where lowering the number of CPUs boosts the DB performance up (concurrency). Point is - Oracle is sweating hard during our tests. You can almost feel the smell of those hot ICs on AWS servers.

And then someone at higher levels, while sitting on a pooper, has a great idea: "I know! Let's migrate to Aurora! They say it's so much faster than anything there is!"

*migration starts*

Tests after migration: the database on the largest instance possible shits itself at 10% of the previous load: the CPU% is maxed out (sy:60%,us:40%), IO is far, far from hitting the limits.

Is it really possible Aurora will cope with the load better than Oracle? Frankly, I haven't seen any database perform better than Oracle yet. Not sure if it's worth to invest time in this adventure..

I'm currently between jobs and have a few rants about my previous job (naturally). In retrospect, it's somewhat therapeutic to range about the sheer brainfuckery that has taken place. Enjoy!

First, let me set the scene: legacy B2B web app made with LEMP stack and sencha ext.js 3 + 4 (don't ask) and a lot of madness. Let's call that app "Alpha".

Alpha is a self made CMS build for typical ERP stuff. Yes, a self made CMS: entities are containers, containers have types and fields and values. Like so many legacy PHP apps, it does not have a dedicated FE: the HTML is rendered on the server and then spewed out to the browser.

Easy right? Coding like it's 1999! But there was a twist: Because everything is basically a container, the HTML-templates are saved in the DB. Along with the nessary JS and the CSS. And the translation variables. Why? Because fuck you! That's why. Who needs a git history anyways.

For some reason, Alpha was kinda slow.

There was also an editor, that allowed you to modify templates (web, mail, pdf) on the fly in prod. Because templates contain repeating data (header/footer), one template could contain additional templates. Much confusion. You could change templates via migration (slow, boring) or just ctrl-c/ctrl-v that sucker (fast, much excitement).

Did I mention Alpha was slow?

On with the rant: e-mails! How do they work? Noone knows. How to send mails asynchronous in PHP? Witchcraft is the only possible answer to that riddle. Here is your enterprise™ solution:
1. create mail
2. insert mail into DB
3. WAIT UP TO 59 SECONDS FOR A FUCKING CRON TO SEND MAIL

Why? "Because that way, we can resend mails in case the network is down :)"

Same procedure for the SOAP-API (db-queue + cron). You read that right: all requests to various other systems are processed once a minute.

Alpha slow.

Alpha was only one of several systems. Imagine a bunch of monolithic php apps, interconnected via SOAP, REST and GraphQL like a godamn intergalactic orgy. Image having to debug that cluster fuck.

Let's say there is a bad request. These things happen. No biggie. Remember the db-queue? Let's try to send the bad request a second time! And a third time! Still no luck? How odd. Let's create a specific file in a specific directory: a LOCK-file. Now, "the db-queue is on hold and no request gets processed :)"

Golly gee thanks Alpha.

Anyhow, did you know that MySQL has a join limit of 61 tables?

Allrighty, so we have a huge migration upcoming. The planning started early this spring. We've split the whole process into separate tasks and estimated each of them. Also marked all the tasks client should take care of itself so save funds and time. All-in-all the whole thing estimated like 4 months if we did it [single dev, tremendous amounts of communication with various parties, buy and prepare the infra, adapt app to the changes, testing, monitoring, etc.] and like a month if client did the tasks we shouldn't be doing. The funding for migration is time-bound and can only be used before December. Cool! We got notified that by the end of April we should be good to go! Plenty of time to do things right!

April comes. Silence. Mid-april we resch out to the client. Since there's plenty of time left migration is getting lower priority to other tasks. Well allright, sort of makes sense. We should migrate mid-July. Cool!

July comes. Client replies that everyone's on vacation now. Gotta wait for August - will do the quicker version of migration to make it on time. Well allright....

August comes. Everyone's vusy with whatever they've postponed during summer. Hopefully we'll start migration in September. Mhm...

September comes. We're invited to a meeting by project funders to explain tasks' breakdown, justify the time needed to make the migration. We're being blamed for surreal estimations and poor organization of tasks as nothing's happened yet... [they were the ones who always were postponing things....]. Moreover, they can only spare 20% of infra resources required for data alone anf they want us to make that enough for all environments, all components, all backups, all databases,... You get the pic.
The leader of the meeting semi silently mumbled to other participants 'Well then I'm afrsid we can't make a full migration in time.. Only partial. That's very unfortunate, very. That's why we should not have incopetent vendors [*glancing at us*]'

somehow we agreed we'll get the resources mid-November and we should be thankful for him bcz he'll have to pull some strings for... us..

I left the meeting with my fists squeezed so hard! But it's okay, we got smth useful: resources and start date. Although it leaves us with less than a month to do smth requiring a month for a sunny-day scenario. Nvm, still doable.

Last week we get an email that resources will be available at the beginning of December [after deadline] and we should start a full migration no sooner than Nov 12. Which leaves us with 50% of our estimated fucking optimistic scenario time and not enough resources to even move a single db.

Fuck I hate politics in dev... Is it wrong for me to want to tie them to a pole, set them on a veeery slow fire and take a piss on them while they're screaming their shitty lungs out? I'd enjoy the view and the scream. I know I would. And while enjoying I might be tempted to take a burning 20cm diameter wooden stick and shove it up their assholes. Repeatedly. Round-robin. Promissing them I'll take it out in 5 seconds and pulling it out after 2 minutes.

Can I?

Mine: not listening to early warning sings that shit won't work. As a result we implemented the wrong technology and had to roll back after a month in production. Lost some customers data, though we had backups, so it was just a delta of a day or so. Still, not pretty, had a rough night writing scripts at 3am to check data validity after.

I will throw in another one from a colleague of mine. He was running some database migrations and ended up pointing the DB migration tool to production by accident. Oops. That one required a restore as well, if I remember correctly.

we're doing a massive database migration and trying to fix a lot of shit that was done for years in the db. problem is, i have to cater to a business major asshat that doesn't know the first thing about working with data and he's responsible for 99% of the shit we're trying to clean up.

his response to the problem i brought up in his stuff? "we can deal with it later, right now i need it like this". this is why you guys have a shit database, because we have to spoil this idiot. then they complain everything takes forever to run and the database is bloated and somehow it is our fault.

I'm really holding myself right now, because i already went off on him once and he basically called me hysterical, and our boss likes him too much to antagonize the bastard. but god i wish i could run over him

Good to see people are learning and don't use hard coded credentials and store them in the DB.
Now they just have to quit putting them into migration files

get to the office at 5:45AM to do a migration because the client doesnt want to automate it.. but pays extra for the inconvenience...

omg fuck you

why did you fucking recreate the fucking db migration on the same fucking version you fucking moron in the fucking develop branch you fucker

sorry it's not my fucking day this piece of shit company with the fucking fake developers
crap, I'm also working in the same company, means that I am also a fucking piece of shit developer

Come on, WordPress! Why are you such an asshole? I just want to migrate and move on with my life.
Is it because I started with Joomla? Is it because I cheated you with Flask?
Can you please, please be nice to the same db you made? Am I asking to much for?

I so hate working with legacy db code! I wish I could migrate all to entity framework but the migration is a pain in the ass and would take hours which the client wouldn't like.

So many choices for backend I'm fucking confused. Yesterday, I tried Django and i found it little similar to RoR(in generating things, db migration things).
I'm currently working in NodeJS.
I even don't know should I rant or cry.
And of course frontend is another thing same like this.....
And I'm not much experienced to differentiate them and know which is better and where it will fit.
Can anyone tell me in simple way which framework fits where?

I have to build a database migration that generates user handles. The user handles are unique within an organization. The user can change them. The auto generated handles are either the first name + last name, or the business name depending on which user type it is. Unless it would be a duplicate. Duplicates auto increment if the handle is taken. The character limit for a user handle is the same length as first name plus last name so I have to check for possible overflow if I add digits. I also have to see if the generated name is in the DB already because a user could have custom entered the result of the auto generation.

This has to be programmed async. The DB driver is using a transaction but multiple calls have to be made to check if the generated handle exists for that organization. Also I have to check the migration script itself for possible duplicates. 3/4 of the users have a handle and with the scale there will definitely be duplicate names.

My idea is if there is a collision, use a UUID and let the users pick something nicer next time they log in. Business says “Reeeeeee!!!! The users shouldn’t see a UUID!!! You can do this!!!” Absurd uniqueness requirements. Absurd backfill procedure. Absurd business rules.

I thought today was a good day to look at how I will deal with database migrations for this node.js/sql-server application. I read up docs for a few migration frameworks but the ones I found seem to make things too complex.

I am tempted to just roll my own by storing a db version in a table, numbering .sql scripts in a folder and running all the higher numbered scripts when the application starts.

Anyone know is there any gold standard for this sort of thing or anything to watch out for?

Guessing my rant free streak is over. Trying to connect to a mongo atlas cluster. Just migrated from mlab as mongo Inc is discontinuing the heroku add on.
Migration went well. I can connect to atlas cluster via mongo shell.
Reactive mongo claims it supports dns seed list. I add mongodb+srv connection string. Doesn't work.
I go back to atlas and allow all ips access (migrating staging dB first to make sure all is well so I can whitelist all ips) - > send a request-> mongo error. No primary node is available.
Disconnect from my network, connect to another network, same thing. I push the connection string to my server, test using an ssl connection to make a request, still no primary node available. I am about to lose my mind.

Why does Microsoft make entity migration default and every tutorial show them? I have other ways of migrating my db schema changed. Newbies come to me with this cluster already in motion and I have to help straighten it out. Then the newbie says "why do all the tutorials show it". I say because the people making the tutorials are trying to quickly create content for a tutorial.

Dont be that guy, ok. Just clean up your shit and don't let shit go through you.
I will git blame you. I will judge you.
#leaveitbetterthanyoufoundit

just 'Hello world' me trying to make a restful api.
*Got Ktor, loved the koltin, hated the deploy, quit.
*Got Django, loved the python, hated the sql migration, quit.
*Got Node, loved everything, hated mongo, can't quit now...
*Got Firebase DB now, I feel the hate monster...ghostly voices, saying, Work my slave, build it... dont stop, 'cause we're right behind you...

....and we're waiting for you

Junior: "I've deleted the unnecessary migration files after a rollback, just like you recommended me, but the DB-table that tracks which migrations where run still shows them after running migrations"
Me: "That shouldn't happen 🤔 How did you delete them?"
J: "I deleted the code in the Up and down scripts"....

Is it so hard to understand that a migration will "run" even if it doesn't do anything?

How to deal with having to work on a very boring task?

I work as android dev in a company where there are around 40 of android devs in total. When I was working on frontend architecture and UI related tasks everything was perfect and I loved my job.

But now for the next couple months all of us have to work on migrating all of our db layer models and business logic related to them to a new built in house ORM library and its a total trainwreck.

Everyone is confused, the task is very boring and obscure and deadline is around 2 months. Just to clarify: one guy did 50% of migration and it took him couple years. Now they are throwing roughly 20 devs at the problem thinking they can do this in 1-2 months.

One week already passed and TBH I havent even started working on this, I just picked up a task and now Im trying to wrap my head around this. Its so boring and obscure and expectations are so unrealistic that I want to kill myself.. Thinking of just taking my 2 weeks vacation to escape this shit or even quitting my job if this goes on longer than a month or two.

me: FE in work, but doing fullstack on my passion projects and somewhat confident on small VPSs - heck, I have a beard, I can do server stuff :) - migrating a WP site that just wont work, copied everything, didn't work, used a migration tool, didn't work, always getting "Connection refused"... must be something with the SSL certificates.. 3 fckn days passed by and nothing when I stumbled upon a forum post with similar issue where the guy stated: I tried all the obvious like copying files, db, certificates, enabled ssl on apache... then it hit me, this is a new installation, I didn't enabled SSL in apache sudo a2enmode ssl restarted apache and BOOM everything is working
part of me was like how stupid you have to be - but the other part is like I guess I learn something every day, this is how you migrate a WP site with the domain #IloveIT

DB migrations give the chills.

This one went well. It worked as expected but that "uneasy" feeling lingers...