Details
-
Bug
-
Resolution: Incomplete
-
Major
-
3.0
-
None
-
Security Level: Public
-
None
-
3.0.0-661-rel
-
Untriaged
-
Centos 64-bit
-
-
Unknown
Description
scenario:
2 clusters, 2 nodes each, 2 buckets, fullEviction policy. Bidirectional xdcr
Loaded initial set of 20000 items
warmed up one source node
warmed up one destination node
update\delete items (about 10000 were deleted)
Actual resutlt:
source cluster has 13993 items in default bucket and destination has 13996
Expected result:
source and destination have same number of items
some of the items are lost in source, but in destination remain all of them:
Source:
172.27.33.10
curr_items: 6994
curr_items_tot: 13993
vb_active_curr_items: 6994
vb_pending_curr_items: 0
vb_replica_curr_items: 6999
172.27.33.11
curr_items: 6999
curr_items_tot: 13996
vb_active_curr_items: 6999
vb_pending_curr_items: 0
vb_replica_curr_items: 6997
Dest:
172.27.33.13
curr_items: 6999
curr_items_tot: 13996
vb_active_curr_items: 6999
vb_pending_curr_items: 0
vb_replica_curr_items: 6997
172.27.33.12
curr_items: 6997
curr_items_tot: 13996
vb_active_curr_items: 6997
vb_pending_curr_items: 0
vb_replica_curr_items: 6999
test:
xdcr.biXDCR.bidirectional.load_with_async_ops_and_joint_sets_with_warmup_master,items=10000,ctopology=chain,rdirection=bidirection,standard_buckets=1,doc-ops=create-delete,doc-ops-dest=create-update-delete,upd=30,del=30,replication_type=xmem,warm=source-destination,GROUP=P0,eviction_policy=fullEviction,dgm_run=true,active_resident_threshold=50
cbcollect will be attached