Hey folks, I'm running into some serious issues with my automated pipeline that performs a Point-In-Time Restore of an Azure SQL database using `Restore-AzSqlDatabase`. For performance reasons, we first restore the database at the P11 tier, export it to a .bacpac, and then delete the restored database. To avoid delays, I have a cleanup task that runs for 40 minutes to check for the database's presence. If it shows up, it's deleted immediately. However, I just got hit with a huge bill for a P11 database that was never usable during the restore process. It turns out the restore took over 6 hours, and even after monitoring, the database didn't show up in the portal or scripts until 6 hours and 40 minutes later! It got charged since it wasn't deleted in time. I raised a support ticket, but Microsoft said no to a refund. What do you all think? Do we lack adequate safeguards here or is this an unfair charge for a backend issue?
4 Answers
Honestly, it seems like this situation is partly on you. You kicked off a restore that should have resulted in a running database. If you set a time limit for cleanup in your pipeline, that's something you need to own. Expect delays on the platform occasionally. To prevent surprises like this, consider setting up budget alerts. Also, think about adding a failure alert in your pipeline for when processes don't complete as expected.
It might be worth considering switching to Hyperscale. You could take snapshot backups and do your exports from there. They won't grant you a refund because you really should have set up better monitoring in your pipeline for the asynchronous restore status. I'm sorry this happened, though! Also, how busy is your source database?
I see what you're saying, but it feels like you're trying to game the system by switching to a higher tier for the restore and then changing it back. That method probably isn't officially supported for refunds when things fail. Just a thought!
I'm feeling conflicted about your situation. On one hand, when you're scripting with expensive resources, you should prepare for all possible issues. If the cleanup didn't finish in 40 minutes, there should definitely be alerts informing you. I think extra notifications could help. That said, I agree—the restore processes can be pretty frustrating. I have a massive bacpac file that's been stuck at 80% for days, and it’s been a bumpy ride overall. It doesn't inspire confidence in the disaster recovery plan unless I’m ready to drop big bucks.
Related Questions
Can't Load PhpMyadmin On After Server Update
Redirect www to non-www in Apache Conf
How To Check If Your SSL Cert Is SHA 1
Windows TrackPad Gestures