search

cityofaustin / atd-data-tech

Austin Transportation Data & Technology Services
17 stars 2 forks source link

Upgrade VZdb to `graphql-engine` 2+ #6988

Closed patrickm02L closed 1 year ago

patrickm02L commented 3 years ago

Upgrading to Hasura 2.0 would be a smart move long term and have added benefits to the Vision Zero Product.

From the conversation in the Dev Backlog Grooming on August 18th we determined that it would be feasible to do, but that it would be tricky because we're dealing with the Terraform script and that recovering from an error is not easy.

It was recommended by John, that in addition to upgrading to Hasura 2.0 the Devs on the team would shadow whomever was taking the lead in order to understand the upgrade process.

This issue came about because it is preventing us from moving forward in a clean way with #6915.

patrickm02L commented 3 years ago

There's a concern about the recovery time. Thinking about that, maybe there's a way to test either locally or in an isolated new instance that would work.

There's some upfront testing that needs to be done: working on the pipeline and integrate migration.

@sergiogcx @mateoclarke @frankhereford would be comfortable working through this. Prep work before Sergio goes on vacation, then deploy when he gets back.

patrickm02L commented 3 years ago

Moving to backlog, so that we can prioritize creating an updated Dev Pipeline (blocker).

Also, it was recommended that we wait until Hasura 2.0 reaches 2.1 so that we know that there's a stable version to move to.

mateoclarke commented 3 years ago

Here is the miro board frame that explains the steps in the process. As of know, we're on step 4 of the process. We need to add seed data that would be initialized when a new local environment when it gets created. Depending on the size of the seed data, we may consider rolling the seeded data into a docker image which would speed up the time required to spin up a new environment.

patrickm02L commented 3 years ago

@patrickm02L Turn this into epic, then use the Miro board to create individual Issues.

mateoclarke commented 2 years ago

Added the Tech Debt label and reassigning to Chia and Frank for later prioritization.

frankhereford commented 2 years ago

Based on the work of #666 for Moped, I'm optimistic that this is going to be a smaller task than we may have thought. We're massively behind in the version of graphql-engine used for VZ, and I am very eager to see that caught up. @patrickm02L @chiaberry - do you see a chance where we can we get to this in a < 60 day timeline?

patrickm02L commented 2 years ago

I'm putting this on the agenda this week to revisit in grooming on 9/12/22.

patrickm02L commented 1 year ago

In Planning 2/22/23:

patrickm02L commented 1 year ago

6/5/23 - Chatted with Frank about potential to close. He responded:

the dev stack is running a modern version of graphl-engine, but I’d be hesitant to close it until there is a finalized set of metadata and a way to apply it easily into production. I think it’d be reasonable to spin that off into its own issue and close this one, or this will be blocked on closing until we freeze the schema of the database just prior to launch.

We'll keep this in QA for now.

frankhereford commented 1 year ago

It's always fun to close these old ones. This issue had been superseded by a more recent one, so please see it for more information. https://github.com/cityofaustin/atd-data-tech/issues/13392