audit super search fields
Categories
(Socorro :: General, task, P2)
Tracking
(Not tracked)
People
(Reporter: willkg, Unassigned)
References
Details
Attachments
(1 file)
As part of taming the processed crash beast, we should do an audit of super search fields and remove things from super_search_fields.py that we aren't using.
- fields that are obsolete
- fields that have been replaced by other fields
- fields that aren't used for search or aggregations
|
Reporter | |
Comment 1•3 years ago
•
|
||
We no longer index any fields where storage_mapping is None, so we should remove all of these.
There are a few where the description field shows up in the UI:
- json_dump
- submitted_timestamp
- telemetry_environment
I think json_dump and telemetry_environment have their own tabs, so I don't think they show up in the UI in a way that uses the description field. submitted_timestamp does, though. I'll add a note to that field and remove the other two.
I'm going to do a pass where I remove most of them except submitted_timestamp.
Note that removing these fields doesn't change what's in the index, what's searchable, and what results are returned. These fields are presently just cruft.
|
|
Reporter | |
Comment 2•3 years ago
|
||
|
|
Reporter | |
Updated•3 years ago
|
|
|
Reporter | |
Comment 3•3 years ago
|
||
|
|
Reporter | |
Comment 4•3 years ago
|
||
Gabriele says we're done with all things flash, so we can remove support for flash annotations and processed crash fields.
|
|
Reporter | |
Comment 5•3 years ago
|
||
Bumping bugs off my queue because I'm not going to get to them any time soon.
|
|
Reporter | |
Comment 6•3 years ago
|
||
Some related work has been done recently:
Bug #1754874 covered removing unused fields that are sourced from the raw crash from supersearch fields. We added a crash_report_keys item and I could use that to figure out how many crash reports had annotations in the raw crash. Then I removed the items from supersearch fields where we weren't getting data anymore.
Bug #1753521 covers changing supersearch fields that are sourced from the raw crash to instead use normalized/validated data from the processed crash. That will improve supersearch data and let us remove some of the normalization/validation code from the indexing code.
I think the next step to do here is to remove anything that's related to flash.
|
|
Reporter | |
Comment 7•2 years ago
|
||
Bug #1787933 covered removing things flash-related.
I think we're all done here.
Description
•