Closed dylang closed 9 years ago
We are in touch with the author of this package about this behaviour. It seems rapid-fire publishing interacts badly with some mirroring solutions (notably npm-fullfat-registry, which we no longer run but other people do) which is causing a ton of repeat downloads. Over the weekend, 1% of downloads were bigben.
We'll hopefully get bigben to stop but we're probably going to need to do something about the mirrors, as this is happening on a smaller scale with any package with a lot of versions.
Oh, fascinating, thanks for sharing the reason! This also makes me feel better that it won't cause problems with our internal npm-skimdb-registry
.
The package bigben was published a few days ago and has been getting a steady increase of daily downloads increasing by about 18,000 daily, with 112,790 downloads yesterday.
https://api.npmjs.org/downloads/range/last-month/bigben
I'm sure 389,314 downloads the first week is possible, but this particular package doesn't do anything. The author is using a script to change the version and re-publish it every 15 minutes.
Could there be a bug that results in downloads to artificially increase when a package is published?
Discussion with the author: https://github.com/OndrejP/NpmjsBigBen/issues/1
(I know
download-counts
is only used to query the stats and not collect them, but I wasn't sure where else to put this ticket. I'm happy to re-create it wherever it belongs.)