Description
1) Loaded data into a 9 node cluster at 181.
2) Stopped load, waited for disk-write-queue = 0.
3) shutdown all nodes simultaneously via /etc/init.d/couchbase-server stop
4) upgrade to 2.0.2-807.
Nodes come back up but with less items than before upgrade. 2 nodes have 0 active items.
Impact: apparent data loss
Cluster live ( 10.3.2.89:8091)
all node diags attached.
shutdown occured at timestamp: [2013-05-20 14:37...]
followed by upgrade.