Details
-
Bug
-
Resolution: Duplicate
-
Critical
-
None
-
6.5.0
-
Untriaged
-
Centos 64-bit
-
-
Unknown
Description
Steps:
On node 6.5.0-3821, durability as MAJORITY
- create a transaction and load 3000 doc
- Add a node in and start the rebalance
- while rebalance is going on, update already loaded 1500 doc.
- All the load completes successfully and the rebalance fails.
test case:
/opt/gradle/*/bin/gradle --refresh-dependencies testrunner -P jython="/opt/jython2.7.0/bin/jython" -P args="-i b/try.ini -t rebalance_new.rebalance_in.RebalanceInTests.test_rebalance_in_with_ops,nodes_init=2,nodes_in=1,replicas=1,standard_buckets=2,sasl_buckets=2,standard_bucket_priority=low:high,sasl_bucket_priority=low:high,doc_ops=update,value_size=1024,transaction_commit=True,GROUP=IN;P0,num_items=3000,atomicity=True,transaction_timeout=3000,bucket_size=100,skip_cleanup=True -m rest"
Attachments
Issue Links
- duplicates
-
MB-35208 Rebalance failed during replica update if disk backfill and memory snapshot contain same Prepare key
- Closed