-
Type: Bug
-
Resolution: Duplicate
-
Priority: Critical - P2
-
None
-
Affects Version/s: 3.0.5
-
Component/s: Replication
-
None
-
ALL
-
slave node crash with the log:
2015-12-05T12:20:45.640+0800 E REPL [repl writer worker 12] writer worker caught exception: :: caused by :: 11000 E11000 duplicate key error collection: data.rs2 index: _id_ dup key: { : "77g9ya:1_390de16e-744f-11e5-8000" } on: { ts: Timestamp 1449289245000|699, h: -2990965226155869115, v: 2, op: "u", ns: "data.rs2", o2: { _id: "77g9ya:1_390de16e-744f-11e5-8000" }, o: { _id: "77g9ya:1_390de16e-744f-11e5-8000", hash: "FuVlaMfDo7sMDmpX6hBLm-jOSSXY", fdel: 1} } 2015-12-05T12:20:45.640+0800 I - [repl writer worker 12] Fatal Assertion 16360 2015-12-05T12:20:45.640+0800 I - [repl writer worker 12] ***aborting after fassert() failure
after restart the slave node, it run normally, 3 hour passed, not crash again. the document in the slave is equal to which in master.
more information:
1. db version: 3.0.5
2. storage engine: wiredTiger
3. replset: 1master + 2slave + 2arbiter (only one slave node crash.)
4. index: only "_id" index, and never change
5. both master and the slave run more than 1 month, never upgrade, never restart, never migrate
6. one day ago adjust wiredTiger config use the command on master node(not sure if this affect the replication):
db.adminCommand({ "setParameter": 1, "wiredTigerEngineRuntimeConfig": "cache_size=25G, eviction_trigger=85" })
log:
2015-12-04T11:24:22.178+0800 I COMMAND [conn1239841] command admin.$cmd command: setParameter { setParameter: 1.0, wiredTigerEngineRuntimeConfig: "cache_size=25G, eviction_trigger=85" } keyUpdates:0 writeConflicts:0 numYields:0 reslen:47 locks:{} 9950ms 2015-12-04T15:01:01.114+0800 I COMMAND [conn1244358] command admin.$cmd command: setParameter { setParameter: 1.0, wiredTigerEngineRuntimeConfig: "cache_size=25G, eviction_trigger=85" } keyUpdates:0 writeConflicts:0 numYields:0 reslen:47 locks:{} 6069ms
- duplicates
-
SERVER-21275 Document not found due to WT commit visibility issue
- Closed