-
Type:
Bug
-
Resolution: Duplicate
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
Catalog and Routing
-
ALL
-
None
-
None
-
None
-
None
-
None
-
None
-
None
On master, the dropIndexes suppresses writeConcern errors.
Successful execution of the dropIndexes command:
1 {
2 "nIndexesWas" : 2,
3 "ok" : 1,
4 "raw" : {
5 "mongos_noop_writes_wait_for_write_concern-rs0/ip-10-122-1-7:20043,ip-10-122-1-7:20044,ip-10-122-1-7:20045" : {
6 "nIndexesWas" : 2,
7 "ok" : 1
8 }
9 },
10 "$clusterTime" : {
11 "clusterTime" : Timestamp(1731848532, 1),
12 "signature" : {
13 "hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
14 "keyId" : NumberLong(0)
15 }
16 },
17 "operationTime" : Timestamp(1731848532, 1)
18 }
Note the writeConcern error in case of a successful execution of createIndexes:
1 {
2 "numIndexesBefore" : 1,
3 "numIndexesAfter" : 2,
4 "createdCollectionAutomatically" : false,
5 "commitQuorum" : "majority",
6 "writeConcernError" : {
7 "code" : 64,
8 "codeName" : "WriteConcernFailed",
9 "errmsg" : "waiting for replication timed out",
10 "errInfo" : {
11 "wtimeout" : true,
12 "writeConcern" : {
13 "w" : 3,
14 "wtimeout" : 1000,
15 "provenance" : "clientSupplied"
16 }
17 }
18 },
In v5.0, on retrying a dropIndexes command (due to some transient error), write concern errors may be suppressed by a write error such as "IndexNotFound", and a user may mistakenly believe that the dropIndexes command was executed successfully on a majority of nodes; when in fact this dropIndexes action can be rolled back.
The response on running a dropIndexes command in v5.0 contains the wCE when the index is first dropped:
1 {
2 "raw" : {
3 "mongos_noop_writes_wait_for_write_concern-rs0/ip-10-122-1-7:20020,ip-10-122-1-7:20021,ip-10-122-1-7:20022" : {
4 "nIndexesWas" : 2,
5 "writeConcernError" : {
6 "code" : 64,
7 "codeName" : "WriteConcernFailed",
8 "errmsg" : "waiting for replication timed out",
9 "errInfo" : {
10 "wtimeout" : true,
11 "writeConcern" : {
12 "w" : 3,
13 "wtimeout" : 1000,
14 "provenance" : "clientSupplied"
15 }
16 }
17 },
18 "ok" : 1
19 }
20 },
On retrying the dropIndexes command:
1 {
2 "raw" : {
3 "mongos_noop_writes_wait_for_write_concern-rs0/ip-10-122-1-7:20020,ip-10-122-1-7:20021,ip-10-122-1-7:20022" : {
4 "ok" : 0,
5 "errmsg" : "can't find index with key: { a: 1.0 }",
6 "code" : 27,
7 "codeName" : "IndexNotFound"
8 }
9 },
10 "code" : 27,
11 "codeName" : "IndexNotFound",
12 "ok" : 0,
13 "errmsg" : "can't find index with key: { a: 1.0 }",
14 "$clusterTime" : {
15 "clusterTime" : Timestamp(1731597701, 1),
16 "signature" : {
17 "hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
18 "keyId" : NumberLong(0)
19 }
20 },
21 "operationTime" : Timestamp(1731597701, 1)
22 }
Response when a createIndexes command (contains the wCE) is retried:
{
2 "raw" : {
3 "mongos_noop_writes_wait_for_write_concern-rs0/ip-10-122-1-7:20020,ip-10-122-1-7:20021,ip-10-122-1-7:20022" : {
4 "numIndexesBefore" : 2,
5 "numIndexesAfter" : 2,
6 "note" : "all indexes already exist",
7 "writeConcernError" : {
8 "code" : 64,
9 "codeName" : "WriteConcernFailed",
10 "errmsg" : "waiting for replication timed out",
11 "errInfo" : {
12 "wtimeout" : true,
13 "writeConcern" : {
14 "w" : 3,
15 "wtimeout" : 1000,
16 "provenance" : "clientSupplied"
17 }
18 }
19 },
20 "ok" : 1
21 }
22 },
23 "writeConcernError" : {
24 "code" : 64,
25 "codeName" : "WriteConcernFailed",
26 "errmsg" : "waiting for replication timed out; Error details: { wtimeout: true, writeConcern: { w: 3, wtimeout: 1000, provenance: \"clientSupplied\" } } at mongos_noop_wri tes_wait_for_write_concern-rs0",
27 "errInfo" : {
28 "wtimeout" : true,
29 "writeConcern" : {
30 "w" : 3,
31 "wtimeout" : 1000,
32 "provenance" : "clientSupplied"
33 }
34 }
35 },
36 "ok" : 1,
37 "$clusterTime" : {
38 "clusterTime" : Timestamp(1731597700, 6),
39 "signature" : {
40 "hash" : BinData(0,"AAAAAAAAAAAAAAAAAAAAAAAAAAA="),
41 "keyId" : NumberLong(0)
42 }
43 },
44 "operationTime" : Timestamp(1731597700, 6)
45 }
- duplicates
-
-
- Needs Scheduling
-
- related to
-
-
- Closed
-
-
-
- Closed
-
-
-
- Closed
-