-
Type:
Improvement
-
Resolution: Fixed
-
Priority:
Minor - P4
-
Affects Version/s: None
-
Component/s: Sharding
-
Labels:
-
Sharding EMEA
-
Fully Compatible
-
Sharding EMEA 2023-04-17, Sharding EMEA 2023-05-15
-
150
The beginning of a range removal is logging min and max values without specifying the shard key field(s).
All other log lines from the range deleter - instead - are also attaching the collection's UUID and redacting the range:
"range"_attr = redact(range.toString()))
Unifying the style (always logging UUID and range) would allow TSEs (or anyone else) to aggregate logs joining the fields without having to perform complex transformations/projections.
That's an example of the described discrepancies:
{"t":{"$date":"2021-11-19T14:47:24.091+00:00"},"s":"D1", "c":"RDELETER", "id":23766, "ctx":"range-deleter","msg":"Begin removal of range","attr":{"min":{"":"1234"},"max":{"":"1235"},"namespace":"db.coll"}}
{"t":{"$date":"2021-11-19T16:05:24.155+00:00"},"s":"D1", "c":"RDELETER", "id":23769, "ctx":"range-deleter","msg":"Deleted documents in pass","attr":{"numDeleted":54381,"namespace":"db.coll","collectionUUID":{"uuid":{"$uuid":"a81a68e7-e11c-424d-bd3f-286e9638e0a7"}},"range":"[{ _id: \"1234\" }, { _id: \"1235\" })"}}