Last night I migrated from the legacy Postgres to the managed Postgres because one of the machines in the old cluster wouldn’t start (and the new replicas weren’t starting either – error 500: dial error…).
So, to avoid headaches and prevent data loss, I decided to pay extra and not take risks with my company. Today, I have a few questions and I hope you guys can help me out
First question is about flyctl: I can’t see my cluster when I run the command fly mpg list (No managed Postgres clusters found in organization XPTO).
Is this expected? Does it take some time to sync?
Second question is about the official documentation: it says I can only host “a single database on that cluster”, but I can see the “Databases” section and I was able to create another one. Is that correct?
In fact, I migrated the database from the old cluster to the one I manually created using that menu.
Third question is about storage: before migrating to the new cluster, I tested everything by creating a test schema, then deleted it after validation. But in the “Overview” menu, the storage used didn’t shrink — in fact, it increased after I re-migrated the official database.
It makes sense that it increased, but I’m not sure if I’ll be charged for that or if it’s just a UI issue.
Not sure if it’s related, but I had to click the delete button multiple times for it to actually remove the test database.
And lastly, backup questions (fourth? fifth? sixth? ):
On the old legacy cluster, I had a nightly routine where I downloaded a .sql file using pg_dump and sent it to Tigris.
Should I keep that routine, or are the fly.io’s automated backups enough?
Thanks for the questions! All good things to help us improve some of our docs/in-app info
It isn’t expected and doesn’t take time to sync. Could you try reauthing your flyctl and let us know if you still have an issue?
fly auth login
You can create multiple databases on an MPG cluster. This is ~brand new so thanks for the reminder to update our documentation on that.
Out of curiosity, had you added data or you just created a new schema then removed it?
At the moment we only bill for provisioned storage and don’t take into account how much is being used. So if you provision a 10GB cluster you’re billed for 10GB. If in the process of using the database you wind up with the volume expanding to 20GB you’re billed for the provisioned usage.
This sounds like a bug, I’ll mention to the team.
This is a personal choice! Though it is one of the things we’re “selling” you by having a Managed Postgres offering
We do a couple styles of backups today. We do a full backup daily, then we also have incremental backups ~hourly. We have internal monitoring and alerting so we’re catching if there are any issues with backups hopefully before you have to.
At the moment, we retain at least backups going back at least 10 calendar days.
They’re stored in object storage so you wouldn’t see them show up in your cluster storage calculation.
I uploaded the full dump from the old cluster for validation. After confirming the process was successful, I deleted the test database (about 20 minutes later).
Got it! Thank you for clarifying that.
~ I had misunderstood the info alert that says: “The prices shown are for a full month of uptime. You’re only billed for the resources you use.”
Great! I’ll turn off the daily routine and keep it weekly instead (I need to sync the staging environment anyway).
):
