Is Fly.io auto upgrading V1 Postgres VMs to V2?

1

I had a staging database and could have sworn it was a V1 app just last week. When I was verifying something today, I noticed it was now a V2 app and that a “Fly Admin Bot” had pushed a release 6 days ago.
image

When I look at the usage for that machine, it only has 5 days logged.

We did nothing to upgrade this VM manually, and I cannot find any documentation or alerts that this was being done.

Is anyone else noticing this? Any official word from Fly?

2

Single instance PG apps are being automatically migrated to Machines. We will be providing a path for HA setups to migrate on their own in the near future. We are working hard to get PG apps off Nomad to address on-going reliability issues.

2 Likes
3

Ah! It’s good to know I’m not going crazy. I was hoping to load-test our staging environment by replicating our HA settings in production. Not sure how the performance will be different between the two types. Is there any good documentation on creating a two-instance leader/replica setup similar to my V1 App in prod from a single-instance V2 machine?

Also, was this communicated somewhere that this migration was happening? I want to ensure I’m subscribed to keep up to date on changes that might affect our production apps.

Thanks!

4

@ckarpyszyn from my experience you just clone a pg machine and they form a cluster automatically. Though as I understand you need 3 machines minimum for quorum on new leader election, in case cluster leader goes down

5

Is there any good documentation on creating a two-instance leader/replica setup similar to my V1 App in prod from a single-instance V2 machine?

Is there any good documentation on creating a two-instance leader/replica setup similar to my V1 App in prod from a single-instance V2 machine?

The fly machines clone command is your best bet at the moment: Horizontal Scaling · Fly Docs

Also, was this communicated somewhere that this migration was happening? I want to ensure I’m subscribed to keep up to date on changes that might affect our production apps.

Communication for single node PG setups has been fairly limited. As long as you’re production PG apps are HA, you will receive direct communication.

1 Like
6

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.