SRE Weekly - SRE Weekly Issue #252

Their on-call started out as four 24 hour shifts per person interspersed throughout the year. Find out how they transitioned to a new approach in a process that spanned the start of the pandemic.

Mary Moore-Simmons — GitHub

A new Meet version had a higher storage usage requirement, and a backend system filled up.

Google

This is webinar on alert fatigue, coming up on January 14.

Sarah Wells — Financial Times

Jamie Dobson — Container Solutions

The chaos experiments you do for security purposes can often expose weak points in reliability as well.

Aaron Rinehart — Verica

Kelly Shortridge — Capsul8

Here are four nifty outside-the-box ideas to use the data you may already have.

Emily Arnott — Blameless

Their custom incident management tool, DropSEV, can detect incident-worthy availability drops and file an incident automatically, obviating the need for an engineer to decide on severity level on the fly.

Joey Beyda and Ross Delinger — DropBox

This one has some additional detail on a November outage involving MySQL replication lag.

Keith Ballinger — GitHub