Details
-
Bug
-
Resolution: Fixed
-
Blocker
-
3.0
-
Security Level: Public
-
None
-
10.6.2.144-10.6.2.150
-
Untriaged
-
Yes
-
June 30 - July 18
Description
1166, 3.0.0
1. Create 1 node cluster
2. Create max buckets allowed (10 buckets)
3. Add 100 K items per buckets
4. Rebalance-out 1 Node and with mutation running
5. Rebalance-in 1 Node and with mutation running
6. Rebalance-out 2 Node and with mutation running
7. Rebalance-in 2 Node and with mutation running
8. Rebalance-out 3 Node and with mutation running
9. Rebalance-in 3 Node and with mutation running
10. Rebalance-out 4 Node and with mutation running
11. Rebalance-in 4 Node and with mutation running
After Step 11 was complete. Was trying to do a rebalance-out operation for 1 node and the rebalance would exit with Bad Replicas issue.
Bad replicators after rebalance:
Missing = [
]
Extras = [] (repeated 1 times)
Test Case::
./testrunner -i ~/run_tests/palm.ini get-cbcollect-info=False,get-logs=False,stop-on-failure=False,get-coredumps=True,force_kill_memached=True,verify_unacked_bytes=True,total_vbuckets=128,std_vbuckets_dist=5,chk_max_items=1000,chk_period=30 -t rebalance.rebalanceinout.RebalanceInOutTests.incremental_rebalance_in_out_with_max_buckets_number,items=10000,default_bucket=False,skip_cleanup=True,vbuckets=128,GROUP=IN_OUT_LONG;P2
This test case will pass but the issue comes after we run this scenario. So its a good way of finding the issue. We tested it for a simple rebalance-out operation without clean up. Note that this test case would fail if we removed skip_cleanup=True parameter as well.
Marking the bug as regression. Wasn't expecting 1 node rebalance-out to fail for bad replicas.