578376 - multiple crashes from a single person should have less weight then many crashes from different people, so that signature rankings are not skewed

Status: RESOLVED WONTFIX

(Socorro :: General, task)

Description

[Wayne Mery (:wsmwk)]

multiple crashes from a single person should have less weight than crashes from different people, so that ranking of signatures is not skewed

hashing of IP address has been suggested as one possible method.

In today's example (but I've seen this many times before) someone is flooding with hundreds of crash reports - sort by the crash date column to see the time clusters https://crash-stats.mozilla.com/report/list?version=Thunderbird%3A3.1&build_id=&query_search=signature&query_type=exact&query=&date=2010-07-13%2005%3A00%3A00&range_value=2&range_unit=weeks&hang_type=any&process_type=any&plugin_field=&plugin_query_type=&plugin_query=&do_query=&signature=qcms_transform_data_rgba_out_lut_sse&missing_sig=&page=1 

The spammage has pushed qcms_transform_data_rgba_out_lut_sse/Bug 578375 to #4 ranking, but it should really be ranked in the #200-#300 range

Also, I'll throw out that it might also be desirable to have some means of identifying these crash bombings, but I don't see a use case.

Comment 1

[(not currently active) Ted Mielczarek]

I thought we had a bug on this, but the closest I could find was bug 529409.

Comment 2

[Wayne Mery (:wsmwk)]

perhaps another example https://crash-stats.mozilla.com/report/list?product=Thunderbird&build_id=&query_search=signature&query_type=exact&query=__delayLoadHelper2&date=07%2F20%2F2010%2012%3A15%3A09&range_value=4&range_unit=weeks&hang_type=any&process_type=all&plugin_field=&plugin_query_type=&plugin_query=&do_query=1&signature=__delayLoadHelper2&missing_sig=&page=1

Comment 3

[Robert Kaiser]

I think the duplicate analysis in bug 629088 and followups helps to some extent here and can form a basis for solving this one - bug 629086 might even overlap with the actual request here.

Comment 4

[Lonnen :lonnen]

There are other means of compaction and modification we may adopt. Crash rates are being moved from socorro to data platform projects like Mission Control and we are unlikely to implement them.