I think I’ve spotted an issue in the DockerHub API related to the “pullCount” in API responses. I’ve detailed it here on Twitter, but here’s the basic issue summarised:
- I’ve been tracking download counts (“pullCount”) from the API for images
- On March 8th between 1700 and 1800 UTC the pullCount for Alpine went from over 2 billion back down to 1.14 billion
- I have been tracking since December and this is the only time a count has gone down instead of up
- This is the only image this has occurred for
- It occurred close to where a typical 32 bit overflow could occur
- It has been increasing normally since it went down
Is this a bug in the DockerHub API? Can the pullCount be relied upon for data scraping and other uses?