Lock wait timeout exceeded after node recovery

[Scott Tully]

I am not sure that my specific setup matters, but i've got 8 data nodes on 4 hosts, 3 api's on seperate hosts, and one management node.

here's the problem/observation:

After a storage node "crashes" and reconnects to the cluster, sometimes a see an error " Lock wait timeout exceeded; try restarting transaction" on one of the API's. This error occurs repeatedly on the same record... and only the same query on that record.

Assumption: this query was executing when the node crashed and is somehow "stuck".

My remedy: restart all the API's. I noticed that just because the error (query) occurs on one API, it doesn't mean that's the same API that is CAUSING this error.

Is there a better way to clear this problem than restarting the API's? Is this a bug?

Thanks,
Scott