Closed Bug 1746850 Opened 3 years ago Closed 3 years ago

Index the DumperError annotation

Categories

(Socorro :: General, task, P2)

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: gsvelto, Assigned: willkg)

References

Details

Attachments

(1 file)

In bug 1666733 we added a new annotation (DumperError) to record failures to write a minidump. This field is currently not indexed so it would be nice for it to be. The field is currently not public and for the time being I think it should stay that way because it's a formatted string (see this list of errors). Once we know what appears in the wild we might want to make it public but only after having sanitized it like we do for raw crash reasons.

I skimmed the related bugs, but didn't see a data review. Is there one?

Flags: needinfo?(gsvelto)

Probably not, I don't remember when I was first told new annotations needed a data review so a few slipped in between when the new policy was put in place and when we started applying it consistently. Should I ask for a retroactive data-review?

Flags: needinfo?(gsvelto)

Yes, please.

At some point (hopefully soon?), I'm planning to go through the crash annotations we're getting and collect metadata for them about what we're collecting, the value domain, why we're collecting it, code, related data reviews, etc similar to what we're doing with data in Telemetry. Then we can make that data easier to search, use, understand for everyone.

I've put the data review request in bug 1666733 comment 5.

Grabbing to do next year.

Assignee: nobody → willkg
Status: NEW → ASSIGNED
Priority: -- → P2

(The new "crash report keys" field is amazing... Yay!)

I looked at some values and they look like unstructured sentences. I think we want to index this like strings rather than like keywords. Then you can do substring searches.

willkg merged PR #5969: "bug 1746850: add DumperError to search" in c116de4.

This will let use queries like:

  • dumper_error exists / dumper_error does not exist
  • dumper_error has terms "somestring" for case-insensitive, analyzed substring searches

This was deployed earlier today in bug #1749582.

We'll need to wait until the weekend for a new index to get created. I'll keep this open and verify then.

New index was created and this field works nicely in search. Marking as FIXED.

Status: ASSIGNED → RESOLVED
Closed: 3 years ago
Resolution: --- → FIXED

Woohoo! This is going to be super-useful

You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: