147 lines
5.1 KiB
Org Mode
147 lines
5.1 KiB
Org Mode
|
:PROPERTIES:
|
|||
|
:ID: 83d8f082-c376-407a-916c-19bbdf3b84e8
|
|||
|
:END:
|
|||
|
#+title: 2021-10-25
|
|||
|
* Discussing Sites removal
|
|||
|
Targetting the complete removal and archival of the sites repo / PHP codebase.
|
|||
|
|
|||
|
- Front-end
|
|||
|
- Back-end
|
|||
|
- Cron jobs
|
|||
|
|
|||
|
** [[id:193f7c04-0a03-4870-90c8-2b5e3c4c92ce][Moving pages out of Sites]]
|
|||
|
*** Required APIs
|
|||
|
- Prefer to update the existing public api vs creating additional services
|
|||
|
*** Move static content out
|
|||
|
*** Breaking apart separate applications/areas
|
|||
|
**** addlead
|
|||
|
**** webforms
|
|||
|
**** control panel
|
|||
|
**** static content
|
|||
|
**** login
|
|||
|
**** order
|
|||
|
**** affiliate
|
|||
|
**** public site (not static)
|
|||
|
**** admin (not an issue)
|
|||
|
**** analytics opens
|
|||
|
**** analytics clicks
|
|||
|
*** Integrations
|
|||
|
** React Shell Application
|
|||
|
- David R is researching "Federated components" in webpack to tie multiple
|
|||
|
smaller react applications together
|
|||
|
*** Move routing from the F5 to the shell application
|
|||
|
- Use www as the base for loading the container HTML / routing?
|
|||
|
- Jim will investigate where we'll go with this
|
|||
|
|
|||
|
** Session & Authentication
|
|||
|
- Replace sessions with OAuth login
|
|||
|
- Identify how to handle authentication and exposure of "hidden" scopes and
|
|||
|
permissions.
|
|||
|
- How do we manage different usage flows?
|
|||
|
- Logged into same account in multiple tabs
|
|||
|
- Different accounts in different tabs?
|
|||
|
- Admin logging in as an account?
|
|||
|
- Investigate abuse vectors
|
|||
|
|
|||
|
* Planning Recipient Service Migration
|
|||
|
|
|||
|
Migrating the recipient service into the production AWS cluster.
|
|||
|
|
|||
|
#+begin_quote
|
|||
|
|
|||
|
|
|||
|
Eric Toner Today at 2:04 PM
|
|||
|
@pse-team Does anyone know of any docs regarding the new AWS cluster? Planning to migrate recipient to the new cluster but I’m not sure what the differences are re: production cluster v. prod-us-east-1-ecs-1
|
|||
|
|
|||
|
|
|||
|
|
|||
|
|
|||
|
20 replies
|
|||
|
|
|||
|
Edward Long:spiral_calendar_pad: 1 hour ago
|
|||
|
(how long will this migration take, will there be any downtime?)
|
|||
|
|
|||
|
Gavin M Roy 1 hour ago
|
|||
|
@Eric Toner I don't know that I documented differences. They are:
|
|||
|
The VPC they reside in
|
|||
|
The version of the AWS Linux distribution
|
|||
|
The name
|
|||
|
The EC2 instance types used
|
|||
|
(edited)
|
|||
|
:thankyou:
|
|||
|
1
|
|||
|
|
|||
|
|
|||
|
Gavin M Roy 1 hour ago
|
|||
|
Tangentially, you should also be using the "newer" ECS base Images, which get rid of the com-aweber-config behavior.
|
|||
|
https://gitlab.aweber.io/PSE/Docker/ecs
|
|||
|
:+1:
|
|||
|
2
|
|||
|
|
|||
|
|
|||
|
Ryan Steele 1 hour ago
|
|||
|
I tried looking for docs on standing up a new service in AWS, but the closest thing I could find in Confluence was https://confluence.aweber.io/display/STD/AWS+Due+Diligence. It seems more high-level in nature though, and the Setting up a New Service page it links to was last modified in 2015 and contains a fair number of broken links.
|
|||
|
|
|||
|
Ryan Steele 1 hour ago
|
|||
|
Wouldn’t hurt to have a canonical source of truth for that info (in a similar vein as our CDE AWS docs).
|
|||
|
|
|||
|
Correl Roush (she/her) 1 hour ago
|
|||
|
I had detailed notes on AWS setup in the bulk tagging playbook that I tend to reference back on, might split that into its own doc as that’s moving into k8s
|
|||
|
|
|||
|
Correl Roush (she/her) 1 hour ago
|
|||
|
https://confluence.aweber.io/display/AR/Bulk+Tagging+Service+Playbook#BulkTaggingServicePlaybook-DeploymentChecklist
|
|||
|
|
|||
|
Gavin M Roy 1 hour ago
|
|||
|
@Ryan Steele we are not adding new services to AWS
|
|||
|
|
|||
|
Gavin M Roy 1 hour ago
|
|||
|
we are moving old services to the new cluster or to k8s
|
|||
|
|
|||
|
Gavin M Roy 1 hour ago
|
|||
|
ideally ECS would be gone.
|
|||
|
|
|||
|
Gavin M Roy 1 hour ago
|
|||
|
sans perhaps the Infrastructure cluster due to what it's done.
|
|||
|
|
|||
|
Ryan Steele 1 hour ago
|
|||
|
Okay, that’s fine. But that documenting our current policy/recommendations would be preferable to having it buried in a Slack thread. (If that information exists already, my Jira search-fu is too weak to find it.) (edited)
|
|||
|
|
|||
|
Gavin M Roy 42 minutes ago
|
|||
|
https://confluence.aweber.io/display/STD/Where+to+Provision+New+Projects
|
|||
|
:docs-props:
|
|||
|
1
|
|||
|
|
|||
|
|
|||
|
Correl Roush (she/her) 30 minutes ago
|
|||
|
@gavinr Is there any documentation on migrating from the old cluster to the new one? (e.g. what needs to get updated, how, and in what order to avoid downtime)
|
|||
|
|
|||
|
Correl Roush (she/her) 28 minutes ago
|
|||
|
My gut says we’ll need to set up an elb, get the service running with it in the new cluster, then update route 53 and remove the old stuff.
|
|||
|
|
|||
|
Gavin M Roy 28 minutes ago
|
|||
|
Not sure, not that I created, but I've walked @dangrosso and others through it multiple times. Am happy to do so with you and Eric when you're ready to do one.
|
|||
|
High Level:
|
|||
|
Using the "New" single stack load balancer cloud-formation template
|
|||
|
New task definitions
|
|||
|
Making sure the project uses the ecs deploy docker image in the CI pipleine
|
|||
|
Deploying the new services to the new clusters
|
|||
|
Moving DNS manually
|
|||
|
Removing the old service definitions from the old clusters once migrated
|
|||
|
Deleting the old Cloudformation stacks
|
|||
|
:thankyou:
|
|||
|
1
|
|||
|
|
|||
|
|
|||
|
Gavin M Roy 27 minutes ago
|
|||
|
The old clusters used one task definition across environments and promoted it on deploy. The new ones use a task-definition per environment and copy that on deployments for that environment only.
|
|||
|
|
|||
|
Correl Roush (she/her) 27 minutes ago
|
|||
|
Ah, that’s good to know.
|
|||
|
|
|||
|
Gavin M Roy 26 minutes ago
|
|||
|
Oh and the newer ECS images work exclusively against SSM Parameter store instead of the old com-aweber-config repos.
|
|||
|
|
|||
|
Gavin M Roy 26 minutes ago
|
|||
|
(though that's independent of cluster)
|
|||
|
#+end_quote
|