Travis CI suffered a prolonged outage this week, leaving users questioning the direction of the continuous integration service following its takeover by Idera in January, and subsequent reported layoffs.
Things started wobbling on Wednesday afternoon UTC, when its status page reported it was investigating reports of “Linux builds not booting promptly”.
By 18.09 UTC on Wednesday it was reporting: “We believe we have identified sources of the issue and we are working on cleaning orphaned VMs that are preventing new ones from being created.”
And 90 minutes later, it was reporting that: “We’ve stopped some workers to mitigate an API rate limit we were running into. This had a positive effect and worker’s capacity is now back up. The queue is still large but we are seeing a downward trend. We’ll keep you posted on the state of the backlog.”
However, it took most of Thursday to get things back up to speed, with the backlog on the Travis-CI.com service for private projects cleared Thursday evening.
Shortly after midnight UTC on Friday, it declared: “We are happy to report that the backlog on travis-ci.org has now cleared as well. Your builds should now be starting more promptly depending on your account’s capacity vs its accumulated backlog. We’ll still be monitoring things closely to ensure the infrastructure continues to be stable. “
At 1.22 UTC it declared the incident resolved. You can see the full incident report here.
However, users were less sanguine, and some quickly made a connection to reports of large-scale layoffs at Travis last month, in the wake of its takeover by private equity group Idera. This led to the creation of a #travisalum hashtag, which was jumped on by other software companies looking to recruit talent.
And inevitably, it has made users question their commitment to the platform.
We’ve asked Idera, Travis’ new parent, to comment on both the outage, and the reported layoffs, but haven’t heard back yet.