[PBM-219] restore fails with "Unrecognized command in op" if new collections created in the oplog Created: 10/Apr/19 Updated: 20/Nov/19 Resolved: 20/May/19 |
|
Status: |
Done |
Project: |
|
Component/s: |
None |
Affects Version/s: |
None |
Fix Version/s: |
Type: |
Bug |
Priority: |
High |
Reporter: |
Assignee: |
||
Resolution: |
Fixed |
Votes: |
0 |
None |
|||
Remaining Estimate: |
0 minutes |
||
Time Spent: |
1 hour, 45 minutes |
||
Original Estimate: |
Not Specified |
Description |
|
Comments |
|
Comment by Carlos Salguero [ 29/Apr/19 ] |
Adamo Tonete What could be the reason of this error? If we are tailing the oplog and then replaying it, why MongoDB says "unrecognized command in op"? |
Comment by Akira Kurogane (Inactive) [ 10/Jul/19 ] |
I'm not sure if this is the UUID problem, or an issue caused by the code not having a method to deal with this sort of oplog record. For interest this is what I got when I ran db.dropDatabase() in a v4.0 instance. It dropped the test collection(s) inside (I didn't request that explicitly) with a "drop" command then ran "dropDatabase" { "ts" : Timestamp(1562733460, 2), "t" : NumberLong(8), "h" : NumberLong("4276643388783434793"), "v" : 2, "op" : "c", "ns" : "fuzz.$cmd", "wall" : ISODate("2019-07-10T04:37:40.805Z"), "o" : { "dropDatabase" : 1 } } { "ts" : Timestamp(1562733460, 1), "t" : NumberLong(8), "h" : NumberLong("-5056640765490911511"), "v" : 2, "op" : "c", "ns" : "fuzz.$cmd", "ui" : UUID("b47184f9-4191-4dbd-8aea-7a2fa3a17f7f"), "wall" : ISODate("2019-07-10T04:37:40.749Z"), "o" : { "drop" : "foo" } } |
Comment by Carlos Salguero [ 10/Jul/19 ] |
This error is fixed. |
Generated at Mon Nov 08 11:57:17 UTC 2021 using Jira 8.13.13#813013-sha1:3ab5fbe7b6c572b1ee4ccec98a9f447334884e00.
Tags: command in, oplogapply command, fails, unrecognized, command, [pbm219], restore