6 ms·
reminds me when I set up an S3 triggered lambda function that also wrote into the same directory. What ensued was millions of files and folders generated recurs
by upupandup 4y ago
reminds me when I set up an S3 triggered lambda function that also wrote into the same directory. What ensued was millions of files and folders generated recursively.
Fortunately, AWS was kind enough to reverse the billing. Had this been Google Cloud, I would've gotten the cold shoulder and a low key threat that if I reverse the transaction I would lose access to my other paid Google products outside GCP :/
- bstpierre 4y agoWith S3 you don't even need lambda to do something like this S3 has some setting where you can log activity on a bucket into another bucket But that setting allows you to set the destination bucket to be the same bucket that you're monitoring. So ~30s after something happens on the monitored bucket, S3 writes a log into the same bucket. And then that activity triggers the logging again. So every ~30-60s, forever, there's a little log written into the bucket. It takes a while to add up to something noticeable if your monthly AWS bill is already a few digits long. It's super fun to sift through the bucket a few months later when you're trying to figure out if there's any real data in the bucket or just endless logs.
- henearkr 4y agoI've had an horror story with Google Cloud, and they have been very helpful. After I explained my situation (a $5k bill for an inactive side/toy project is extremely painful), and I provided great details about what I think happened to cause it, they wrote off the charges.