How can Atlas be a stable or viable solution for any project when it forces random and breaking version changes on you overnight?

Out of nowhere, our users let us know that a significant part of our website had stopped working overnight. After a lot of unscheduled 3am work, we found this was because M0, M2 and M5 clusters were upgraded to a new MongoDB version without any warning or agreement from us. After talking to support, they informed us that automatic version upgrades happened to clusters without warning and these could not be disabled without upgrading to a dedicated cluster (+720% more dollar).

These MongoDB version upgrades can break your app or website overnight, and means Atlas is no longer a viable solution for us. Without bringing mLab into it - but, you know, bringing mLab into it - we never had such issues on mLab.

We did get a $10 coupon from support and they pointed us to these forums, so yeah let me know what you reckon.

1 Like

Hi Daniel,

I very much regret that this happened to you. We clearly need to improve our communications process and evolve our product toward no more breaking changes period. I want to re-earn your trust even if it will be a long road to do so.

You should have received numerous heads-up emails about the upgrade of the Atlas shared tier (M0/M2/M5) clusters to MongoDB 4.4. We will be in touch to investigate how that didn’t happen.

It’s important to emphasize that the vast majority of applications do not rely on capability in MongoDB 4.2 that changed in 4.4: nevertheless it pains us that there are exceptions and we will make major investments in future to reduce the chance of that happening.