The issue is observed as resolved, performance has returned to normal.

We've released a fix, all workspace starts will be done now using the new generation of clusters. The new clusters mitigate the performance issue. We are actively monitoring.

We're steadily shifting traffic from us39a to us40, and eu39b to eu40. These new clusters better handle load and provide improved performance.

Also, we're building a new set of clusters to replace eu39bxl and us39xl, more updates to follow.

This incident is still ongoing. Users are experiencing degraded performance during peak load times. Some (including Gitpodders) have even experienced data loss, for which we are terribly sorry. We are testing fixes internally, and will deploy when confident. Additionally, we will share a transparent blog post to help rebuild your trust.

We've isolated the problem, and are testing a fix.