Data Loss Loading 1 Million Docs to ES Cluster

Used mcsoda to load 1 million documents.
stop mcsoda at end of doc loading,
664 items are missing from elastic search. It's been 2 hours since last data mutation and the missing documents have not been pushed to elastic search.

Still I'm not sure what cause may be. In the couchbase logs there was an error replicating vbucket 802:

2013-01-17 10:47:15 - Error replicating vbucket 802: {badarg, [

{erlang, list_to_integer, ["004340123bf52ca200000HJ5IjoiM2UiLAogInJlYWxtIjoiZWYiLAogImNvaW5zIjozMDcuMjgsCiAiY2F0ZWdvcnkiOjEsCiAiYWNoaWV2ZW1lbnRzIjpbNjAsIDE4NiwgNDAsIDE5Nl0sImJvZHkiOiJ4YzRjYTQyMzhhMGI5MjM4MjBkY2M1MDlhNmY3NTg0OWIifQ==", 16]}

,

{couch_doc, parse_revid,1}

,

{couch_doc,parse_rev,1}

,

{couch_api_wrap, '-get_missing_revs/2-fun-0-', 1}

,

{lists,map,2}, {lists,map,2}

,

{couch_api_wrap, '-get_missing_revs/2-fun-1-', 3}

,

{xdc_vbucket_rep_worker, find_missing,2}

]}

I did a couchdb_dump on this vbucket to see if all it's keys were in the index and they were. I've attached this stat log with this error and will continue to investigate, Marty perhaps you can repro. No errors reported in couchbase logs.