How much space should I reserve for drone’s S3-compatible bucket?
How much space should I reserve for drone’s sqlite db?
Can anyone provide stats on how the space usage correlates with the number of pipelines+steps the drone cluster is running? I suspect that the logs stored in S3 vary widely based on the output logged from pipeline steps. But perhaps each pipeline and each pipeline step requires xxxx KB or xxxx MB in the sqlite db?
Or perhaps there is some rule of thumb like “start with 1 GiB of space and then increase the space based on your usage”. What is a good starting point for bucket and db?
As you mention, this can be difficult to answer because the storage size can vary by build. The clone step for example outputs 500 bytes of logs, however, a single build step can generate up to 5mb of logs before being truncated.
The best I can do is provide some data about our cloud.drone.io instance, which supports around 15,000 developers and 25,000 repositories. The postgres database for cloud.drone.io is a db.t3.medium instance and uses around 3.5GB of disk space, which is pretty reasonable given the number of users and repositories that we support.
I also ran some quick calculations against our S3 bucket. I estimate an average of 1mb of logs per build (which includes all build steps) which means 500,000 builds ≈ 500GB of storage.