Really great communication Dan, thanks for sharing. This really helps drive home the point around metrics, without the ability to see the trees, you'd have had no chance finding the wood!
It's nice to see the transparency and technical detail, but the frequency and duration of outages indicates a more systemic problem to me. I'd want to look at how the Bitbucket team do capacity management, change management, incident management AND problem management.
Excluding Atlassian Access, we've seen 24 incidents in the last three months. Not counting the ongoing one, there has been 40 hours of degraded performance or features. Just barely over 98% availability.
Wouldn't Smart Mirrors (which have been deprecated) help reduce the load? Or was the problem that they were generating too much load during sync? We were really counting on this feature and only found out that they had been deprecated by going to the documentation page.
Apart from one of these blog posts , it appears to be a constant radio silence from bitbucket team on chronic performance issue, we have been bitbucket users for over 6 years but never felt we have to change platform.
Sadly, it appears our development has to move out from bitbucket to more robust git cloud implementations. This load is not suitable for any production scalable environment.
Appears bitbucket is an orphan product of overall Atlassian group. Great aussie business with a side project ( bitbucket )
33 comments