Details
-
Bug
-
Resolution: Fixed
-
Major
-
None
-
2.0-beta
-
Security Level: Public
-
eperf cluster testing on ec2
Description
When running eperf test on a ec2 clust, we see this occasionally in the server logs:
[ns_server:info] [2011-12-16 19:38:29] [ns_1@10.68.154.26:<0.848.0>:ns_port_server:log:155] memcached<0.848.0>: Rubbish received on the backend stream. closing it
memcached<0.848.0>: Trying to connect to mccouch: "localhost:11213"
memcached<0.848.0>: Connected to mccouch: "localhost:11213"
This means somehow the the protocol between mccouch and mc-kvstore is getting mixed up. Fortunately, it looks like no data is lost as the connection is being reset and the data is being resent, but there is some sort of bug that needs to be fixed as it's a performance problem and potentially could be a more serious bug undiagnosed.