That creates a downstream effect of pages taking very long to load, compounding the problem. Anywho, I've just updated all of the configurations to work the way that we expect. Thank you!

And when it does that, the only time it can add more information is when something expires (every cache item has an expiration, usually 60 seconds). So it just has to sit on your request until it get put data into the cache, or eventually gives up and times out.

During the migration, we moved to a new set of servers for running Redis, and they didn't get the same configuration as our old servers. We run Redis fully in memory, and it's meant to eject keys that are old/haven't been looked up in awhile. It was not doing that 💀 So it filled it VERY QUICKLY

We use a database system called Redis for caching information on all of our sites, to make it faster than getting from the database. You give something a "key" (e.g. "here are all of user 1's pets) and then we can easily pull that information later. It's great!

I cannot believe I used to do livestreams during this

The data is all in! Now it's on to phase 3, which is adding those indexes I was talking about.

This tells the database how to find the things we ask for most often FAST

(if you're in the bay area I hope you're staying safe & dry tonight, it's a big storm!)

We can come back up without the activity logs fully in there, unfortunately I dont' think you'd all let me bring the site back up without all of the wardrobe items fully intact 😂

SO CLOSE! 2 tables remain. all of your wardrobe items, and your "activity logs" (the backend logs that we use to track every sP taken, item given, etc)

as you can see 😡

this is where you start to get in to trouble, when you trust the computer and the lies it peddles to you. it has been 11 minutes left for an hour.

multiply that by 20 years! then, we have to do a process called "indexing", so that it's easy to look things up by userid, username, etc. That has to go and re-scan all of the data, and apply the appropriate indexes to it.

there are a LOT of different factors going on here starting with different versions of everything. The different mysql (our database) versions export their data (and then expect the input data) in SLIGHTLY different ways, and don't error until you've made it 85% through a parallelized import

🛁 on the bath house - i wrote a script that gives every a check-in for the last two days (and i'll extend it more!) if they missed the check-in and had recently

so: i'm not going to give another time estimate because I've been wrong every time, but it's still going!