View Issue Details

IDProjectCategoryView StatusLast Update
5730Composr website (compo.sr)General / Uncategorisedpublic2024-04-28 23:36
ReporterPDStig Assigned ToPDStig  
PrioritynormalSeverityfeature 
Status resolvedResolutionfixed 
Summary5730: Telemetry: use hashes for detecting repeat error reports instead of the error itself
DescriptionWhen a site submits an error report through telemetry, we want to reduce the number of entries we have on the site's end especially when the same error is reported multiple times. Currently, this is done by checking if the error text already exists in the database. But this has a few problems:

* The error often has temporary / unique IDs in them, so the same error may not actually match due to a different unique ID
* Same as above but for timestamps or other IDs

Instead, add an additional column in the reported errors for storing an MD5 hash. The MD5 hash is generated using the following process:
1) Start with the error message reported
2) Remove everything that looks like a uniqid or md5 hash (replace with "(uniqid)" or "(md5 hash)")
3) Remove everything that looks like it could be a timestamp (for our intents and purposes, we will define this as any epoch integer between January 1, 2017 [the year v10 came out] and the maximum 32-bit integer) (replace with "(timestamp?)")
4) Run a password censor on it
5) Generate the MD5 from the resulting string, but we still store the unmodified error message string in the database
TagsRoadmap: v11
Attach Tags
Time estimation (hours)
Sponsorship open

Sponsor

Date Added Member Amount Sponsored

Activities

There are no notes attached to this issue.

Issue History

Date Modified Username Field Change
2024-04-24 18:29 PDStig New Issue
2024-04-24 18:29 PDStig Status Not Assigned => Assigned
2024-04-24 18:29 PDStig Assigned To => user4172
2024-04-24 18:29 PDStig Tag Attached: Roadmap: v11
2024-04-28 23:36 PDStig Status Assigned => Resolved
2024-04-28 23:36 PDStig Resolution open => fixed