Page MenuHomePhabricator

Filter out EventLogging data with bunk user-agents
Closed, ResolvedPublic

Description

We are receiving alarms like:
Throughput of EventLogging EventError events is CRITICAL
because an external wiki is sending bursts of malformed events.
The user agent used is "Fuzz Faster U Fool v1.2.0-git".
Those events are already filtered out at Refine time, so do not represent a danger to data ingetrity.
However, they are failing verification and forwarded to EventLogging_EventError topic, which generates the mentioned alarms.

Event Timeline

razzi moved this task from Incoming to Operational Excellence on the Analytics board.
razzi added a project: Analytics-Kanban.
razzi subscribed.

Temporarily going to block this user agent; hoping to deprecate this system eventually.

Change 636493 had a related patch set uploaded (by Ottomata; owner: Ottomata):
[operations/puppet@production] Exclude predefined user agents from eventlogging data

https://gerrit.wikimedia.org/r/636493

Ottomata renamed this task from Filter non-mediawiki hostnames at ingestion time to Filter out EventLogging data with bunk user-agents.Oct 26 2020, 8:17 PM

Change 636493 merged by Ottomata:
[operations/puppet@production] Exclude predefined user agents from eventlogging data

https://gerrit.wikimedia.org/r/636493

Mentioned in SAL (#wikimedia-analytics) [2020-10-27T17:38:00Z] <ottomata> restrict Fuzz Faster U Fool user agents from submittnig eventlogging legacy systemd data - T266130