---
id: qoqpl5nlxw61z2ivy8n1izz
title: Async Work In TFE
desc: Asynchronous Work in TFE Book Club
updated: 1662669501984
created: 1662665156463
traitIds:
- meetingNote
---

# Asynchronous Work in TFE Book Club

<style>
body {
font-family: 'Lato', sans-serif;
}
</style>

## Attendees

- @theo
- @robert
- @sudomateo
- @annie
- @polyglotdev

### RFC to reference

[Async Work In TFE](https://docs.google.com/document/d/1VegEj2fhpsysckfCIxFXgUX-QqN1sh9WnyE3Wxxosdo/edit)

### Agreed upon changes to RFC/Other Questions

- [ ] @annie: [questions about schema verbiage and meaning](https://docs.google.com/document/d/1VegEj2fhpsysckfCIxFXgUX-QqN1sh9WnyE3Wxxosdo/edit?disco=AAAAfq80EII)
> here the agreement is to change the verbiage around 'single shared database' to '**single shared schema**'
- [ ] @theo: asked about - TFEClient(http client) class and sidekick worker(ruby class that adheres to specific class)
> My assumption here is that there is a monkey-patched class that is used to make the http requests to the TFE API
- [ ] @theo: Where are the migrations kept?
> Template in PTFE-Replicated: logic can be put into single binary
- [ ] @theo: in active-active is each node TFE-task-worker select on last 20 and process those?
> Doing this in Postgres requires one query as opposed to Redis and being bogged down with things like locking. So if something happened to power in database with Redis the job is lost but with Postgres it is not because its not being stored in memory or a cache that is more volatile in this use case.
- [ ] @theo: File based config app, but then at one point can we change them on the fly while the app is running? What is that scenario really?
> Changed concurrency for the task worker and sighup to reload the config
- [ ] @sudomateo: making changes to configuration syntax and adding sudo code to appendix.
- [ ] @theo: Are we making sidekick changes with reference to [545](https://docs.google.com/document/d/1lUyt6FjFVhCxipM_YALTV3Xbkn56SOjGyH9oOqaxM_8/edit)
> No, we are not making changes to the sidekick
- [ ] @theo: wants to break down Jira tickets into smaller units of work tomorrow