-
Epic
-
Resolution: Done
-
Normal
-
None
-
None
-
Upgrade Quickstarts Service databases' to 16.4
-
False
-
-
False
-
Unset
-
In Progress
-
-
-
Plat-Ex Services Sprint 47
Description
Defines the work to be done to upgrade the Chrome Service backend PRODUCTION database instance type to Graviton.
Why
Graviton instance are cheaper and provide much better performance. As an example, upgrading a `db.m5.2xlarge` Multi-AZ enabled instance to `db.m6g.2xlarge` will save about $110/month while providing better performance.
Chosen upgrade path
- Chrome service PROD is currently running on this instance_class: db.m5.xlarge
- Upgrading instance_class from: db.m5.xlarge --> db.m6g.xlarge
Useful resources
Announcements to be made for the production upgrade
| Ticket | Task | Email thread/Merge request | Is it done? Is it merged? |
|---|---|---|---|
| N/A | Announce the upgrade via email two weeks in advance | ||
| Set up a Pendo preemptive alert announcing the upgrade in two weeks and schedule another one for the day of the upgrade | |||
| Set up the maintenance announcement in the status page | |||
| Send a preemptive Slack message to the relevant channels | |||
| Send a Slack message to the relevant channels once the maintenance is taking place | |||
| Send a Slack message to the relevant channels once the maintenance is over |
Work to be done by engineers for the production upgrade
| Ticket | Order | Task | MRs |
|---|---|---|---|
| Before anything else | Check that no replication slots are being used. | ||
| Any time before the upgrade | Enable the Blue/Green upgrade settings | ||
| 1 | Specify the new database settings for instance_class to db.m6g.xlarge | ||
| 2 | Wait for the database to be upgraded. | ||
| 3 | Run the post-upgrade maintenance scripts in the database. | ||
| 4 | Clean up the old database configuration files. |
- clones
-
-
- Closed
-