-
Type:
Improvement
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: None
-
Storage Engines
-
5
-
StorEng - Defined Pipeline
While debugging something I turned on verbose=(rts) on a database that has two tables. When running RTS during recovery, I get something like 50+ identical lines that don't really give any information. They look like:
[1638996396:428921][18778:0x7f3b317fa700], t, txn rollback_to_stable: [WT_VERB_RECOVERY][DEBUG]: rollback to stable history store cleanup of update with start timestamp: (0, 0) [1638996396:428926][18778:0x7f3b317fa700], t, txn rollback_to_stable: [WT_VERB_RECOVERY][DEBUG]: rollback to stable history store cleanup of update with start timestamp: (0, 0) [1638996396:428931][18778:0x7f3b317fa700], t, txn rollback_to_stable: [WT_VERB_RECOVERY][DEBUG]: rollback to stable history store cleanup of update with start timestamp: (0, 0) [1638996396:428935][18778:0x7f3b317fa700], t, txn rollback_to_stable: [WT_VERB_RECOVERY][DEBUG]: rollback to stable history store cleanup of update with start timestamp: (0, 0) [1638996396:428940][18778:0x7f3b317fa700], t, txn rollback_to_stable: [WT_VERB_RECOVERY][DEBUG]: rollback to stable history store cleanup of update with start timestamp: (0, 0)
It seems that any cleanup in the HS of an update will print one of these lines. Given that it is debugging level verbose, someone who more regularly debugs history store and timestamp issues should review it and decide if:
- the message is useful as-is, given its vast quantity output
- or if there is any other information that could be given that might help someone know more about this
- or if the message should go away or change form.