-------- Forwarded Message --------
- stream retention change
Date: Mon, 21 Mar 2022 09:00:59 -0400
From: Andrew Otto <otto(a)wikimedia.org>
tl;dr: all publicly available event streams at stream.wikimedia.org
<http://stream.wikimedia.org> will have their retention time set to 7 days.
Many of the streams available at stream.wikimedia.org
<http://stream.wikimedia.org/>have retention times of 31 days. This
means that at any given time, the past 31 days of these streams are
Sometimes, within these streams, certain data may accidentally contain
personally identifiable information. For example, someone might
accidentally enter their personal email into a revision comment field.
On the wikis, this information can be quickly suppressed so that it is
not viewable externally. However, because streams are historical and
immutable, it is difficult to remove this information from the stream
To help mitigate the risk of PII exposure, we are reducing the retention
of these streams to 7 days. We plan to make this change on *Monday April
In the future, we would like to intentionally remove this data from
streams. Doing so requires us to maintain new services that produce new
streams with PII information redacted. Doing this is not a trivial thing
to stand up, hence this mitigation effort for now.