Murphy’s Law dictates that, even in the best-designed data stack, something will eventually go wrong. Queries time out, builds fail, and APIs sometimes go down. Historically, the main challenge to these errors has been knowing when they happen—and why—so that you can diagnose, take corrective action, and get back to work.
Fortunately, Cube Cloud has you covered.
Among Cube Cloud’s valuable features is that you don’t need to read logs to diagnose errors. Instead, Cube Cloud has always included UI tools for managing pre-aggregations and tracing queries.
Now, to further enhance life in the data pipeline, we’re adding another way to make diagnostics easier: configurable email alerts.
With timely notifications when something requires your attention, you’ll be able to triage issues faster than you can say “Kubernetes pod” three times in a row—and before they impact your users.
Four Configurable Alerts
Cube Cloud Alerts take the guesswork out of understanding what’s gone wrong with a deployment. Once they’re set up, the next time something errors out, you’ll receive one of these messages:
“API Failure on Cube Cloud”
API outages happen when a deployment’s API becomes unavailable and an API cannot serve requests. This can happen for a number of reasons, including:
- A bug in the schema code or incorrect deployment
- A database authentication failure
- Internal issues with the Cube Cloud platform
If an API failure occurs, you’ll receive an email alert that includes the HTTP error code of what went wrong:
You’ll then be able to log into Cube Cloud and check for any changes that might have caused the error. And if the cause isn’t super apparent—reach out to us via Intercom for some troubleshooting help.
“Response Time-Outs on Cube Cloud”
API/Database response timeouts are triggered if your APIs or database take longer than expected to respond. It’s helpful to receive a notification when this occurs to alert you to what would otherwise be a silent failure caused, for example, by:
- Large requests that are taking too long to retrieve from the database
- Pre-aggregation build configuration errors
If your APIs or database time out, you’ll receive an alert that specifies the failed SQL query. From there, you’ll be able to log into Cube and trace what went wrong and where.
“Pre-aggregation Build Failure on Cube Cloud”
During a pre-aggregation build failure, data doesn’t load into the Cube Store warmup cache. Unfortunately, like API/database timeouts, these errors can also be silent—making their root causes hard to pinpoint.
A pre-aggregation build failure can be caused by:
- Incorrectly configured pre-aggregations that take too long to build or load
- Configurations that limit simultaneous open sessions and requests to a database
- Overloaded refresh workers or refresh workers that aren’t configured to run in parallel
If a pre-aggregation fails to build, you’ll get an alert with a link to it in Cube. Follow the link to the failed pre-aggregation to see the issue’s root cause and fix it.
Pro Tip. To avoid pre-aggregation build failures, you can configure partitions, export buckets, data source constraints, and refresh workers to keep your cache running smoothly.
Lastly, we’ve added an alert to notify you of error resolutions for all three types of the errors we’ve described here. You’ll get an email that lets you know the problem is fixed so you can continue on your merry way.
Alerts are available to all paid Cube Cloud accounts, and turning them is actually quite simple: just follow the docs here.
Interested in harnessing the power of Cube headless business intelligence in a fully managed deployment? Sign up for Cube Cloud today.