Operations | Monitoring | ITSM | DevOps | Cloud

On Tuesday, Feb. 18, 2025, we experienced an outage that lasted approximately 150 minutes and impacted roughly 25% of our Grafana Cloud services. To our customers: we are very sorry and more than a little embarrassed that we stepped outside our own processes and advice to cause this. You rely on us to help monitor and troubleshoot your environments, and this type of incident obviously makes it harder for you to do that.

Observability is critical for web operations to ensure that the application is working as expected and to identify any potential issues. However, setting up observability has traditionally been challenging because it can take hours to set up all the infrastructure, instrument your code and enable observability in production. But now there is a better way using native support for Django in Charmcraft and Rockcraft which has observability built in and ready to go!

If you're using Playwright for end-to-end testing or synthetic monitoring with Checkly, you've likely considered reusing your test code across different test cases. A common approach for this is using Page Object Models (POMs). However, if you're like me, you might have mixed feelings about POMs—while they help organize your code, they can sometimes feel cumbersome to set up and maintain.

Technical debt accumulates when short-term solutions lead to long-term software inefficiencies, increasing maintenance costs, slowing development, and degrading performance. To effectively manage technical debt, teams need full-stack observability, from a high-level application view down to code execution and thread-level analysis. Tackling technical debt ensures long-term software sustainability.

I’ve increasingly changed my perspective on LLMs and modern AI systems over the past few years: Let me elaborate on why I believe this now.

Believe it or not, we are in the middle of one of the biggest cloud repatriation movements of the past decade. More than ever, companies are rushing to find infrastructure solutions that better suit their needs. Over the past decade, hyperscalers have dominated the market, generating trust and, in some cases, overconfidence in software development. Drawn in by promises of reliability, ease of use, and ultimate flexibility, teams turned to providers like AWS, GCP, and Azure.

Ever wrapped up an incident, thought 'Phew, glad that’s over,' only to feel your stomach drop when you see the dreaded "Incident Debrief" on your calendar? We've all been there. Incident debriefs don't need to feel like sitting through your least favorite school subject. They can (and should!) actually be engaging and useful. At incident.io, we've found a simple, repeatable, and blameless framework.

Atlassian recently announced they’ll be shutting down Opsgenie, their popular on-call alerting tool. After June 4, 2025, no new Opsgenie accounts will be created, and by April 5, 2027, the service will shut down completely. Users don’t seem happy about it. If you’re currently using Opsgenie, this news is significant. A key part of your incident response process is disappearing, and Atlassian suggests moving to their other products, like Jira Service Management or Compass.