Details
-
Type:
Bug
-
Status: Closed
-
Priority:
Major - P3
-
Resolution: Fixed
-
Affects Version/s: None
-
Fix Version/s: 5.0.0-rc0
-
Component/s: None
-
Labels:None
-
Backwards Compatibility:Fully Compatible
-
Operating System:ALL
-
Sprint:Execution Team 2021-05-17
-
Linked BF Score:168
Description
When running a command inserting a batch of time-series documents resulting in multiple buckets, multiple oplog entries will be generated. With {ordered: true} option, currently the field "prevOpTime" in each oplog entry will be set to the same zero timestamp instead of the timestamp of the previous operation.
In the test when setting the command with {ordered: true}, the oplog entries in donorPrimary.getDB("local").oplog.rs look like
{
|
"lsid" : {
|
"id" : UUID("ff76a7f5-239c-4c67-a6ef-d6930899ef5a"),
|
"uid" : BinData(0,"47DEQpj8HBSa+/TImW+5JCeuQeRkm5NMpJWZG3hSuFU=")
|
},
|
"txnNumber" : NumberLong(0),
|
"op" : "i",
|
...,
|
"ts" : Timestamp(1619470064, 5),
|
...,
|
"stmtId" : [
|
0,
|
1,
|
2
|
],
|
"prevOpTime" : {
|
"ts" : Timestamp(0, 0),
|
"t" : NumberLong(-1)
|
}
|
},
|
{
|
"lsid" : {
|
"id" : UUID("ff76a7f5-239c-4c67-a6ef-d6930899ef5a"),
|
"uid" : BinData(0,"47DEQpj8HBSa+/TImW+5JCeuQeRkm5NMpJWZG3hSuFU=")
|
},
|
"txnNumber" : NumberLong(0),
|
"op" : "u",
|
...,
|
"ts" : Timestamp(1619470064, 6),
|
...,
|
"stmtId" : [
|
3,
|
4,
|
5
|
],
|
"prevOpTime" : {
|
"ts" : Timestamp(0, 0),
|
"t" : NumberLong(-1)
|
}
|
},
|
The field "prevOpTime" in the second oplog entry should be the field "ts" in the first oplog entry.
Attachments
Issue Links
- is related to
-
SERVER-55268 Add jstests for tenant migrations with retryable time-series inserts
-
- Closed
-
-
SERVER-54816 Improve performance for time-series inserts with {ordered: true}
-
- Closed
-