| Michael | |
| Aug 5 2025, 3:30 PM |
| F65751852: Screenshot 2025-08-14 at 1.36.48 PM.png | |
| Aug 14 2025, 8:37 PM |
In the client-errors-logstash, I'm seeing a lot of errors (10K per 30 min) for saying RangeError: Maximum call stack size exceeded. (example).
Looking at the user agents, they seem to be all from iOS, including the latest (18.6) and upcoming (26) versions of iOS.
There seem to be no common Gadgets to all the errors. (In particular there are still a couple 100 errors / 30 minutes without the switcher default gadget.)
It started on 25th (wmf.6)
The stack trace is undefined so this seems very suspicious - usually when this happens it is a bot. It looks like all errors are from 20 different IP addresses which support that theory.
Note: we currently ignore stack trace undefined in our alerts so perhaps we should also remove these from the dashboard? There is little we can do to debug such errors.
The current rate of error is quite high - https://logstash.wikimedia.org/goto/791e544ca37a8729620c6a0af47164ad. And, yes, several "undefined" are in the log reports:
error_context.component: unknown
file_url: undefined
stack_trace: undefined
When file_url and stack_trace are undefined this is usually an indication of bot activity. We filter those fields by default from the dashboard and alerts.
This is still ongoing. I tried to reproduce on a real iPhone (Safari, Chrome iOS, and GSA) and on the iOS Simulator but could not trigger it. Likely bot traffic as Jdlrobson noted.