I should write a blog post about this but the TL;DR is that older berkshelf-api servers had to download every cookbook to build the /universe endpoint. They, by default, did this every 5 mins.
I will try and get a post together next week with more details. I do want to make it clear that, until supermarket was released, downloading every cookbook was the only way to get the data needed to build the /universe endpoint.
In about two weeks an empty cookbook https://supermarket.chef.io/cookbooks/empty have around 224,000 downloads
I definitely think there is something wrong in the download counter.
In about two weeks an empty cookbook https://supermarket.chef.io/cookbooks/empty have around 224,000 downloads
I definitely think there is something wrong in the download counter.
I actually wrote about this recently and my response to the original never posted to the list because, apparently, email is hard.
Not only did your informative response of January 18 which explained
what is happening with the download counter get posted to the list and
distributed to subscribers (to me, at least), but Vincenzo quoted your
response (and Julian's, and Noah's) in his followup post today.
This list might benefit from an FAQ and a welcome message for newbies.
E-mail is not hard. We solved most if not all of the communications
problems seen regularly on this list decades ago on other mailing lists
and on Usenet newsgroups. There are good reasons for providing
informative subject headers, asking questions "the smart way", trimming
your posts, and refraining from TOFU quoting.
Thanks Christopher for your answer, nevertheless this counter is
misleading, something should be done.
If someone doesn't know the story behind the download counter, for sure
think: this cookbook should be almost perfect, if there are so many
downloads.
I spent 2 days before I understood that a cookbook with about 16 million
downloads was heavily bugged.
Instead to wait that berkshelf-api servers to run more efficiently, the
rules behind that download counter should change.
May be split the number of download done by a person from the number of
automatic downloads (HTTP User Agents).
Thanks Christopher for your answer, nevertheless this counter is
misleading, something should be done.
If someone doesn't know the story behind the download counter, for sure
think: this cookbook should be almost perfect, if there are so many
downloads.
I spent 2 days before I understood that a cookbook with about 16 million
downloads was heavily bugged.
Instead to wait that berkshelf-api servers to run more efficiently, the
rules behind that download counter should change.
May be split the number of download done by a person from the number of
automatic downloads (HTTP User Agents).