Details
-
Bug
-
Resolution: Unresolved
-
Critical
-
3.0
-
Security Level: Public
-
None
-
Untriaged
-
No
Description
There's a chance that memcached requests that are lost (from dead OS or erlang process for example or timed out) are still going to be executed by memcached. Causing arbitrary issues.
There's agreement with ep-engine and memcached folks and working implementation of new memcached extension described here: https://docs.google.com/document/d/116_7evfelOL5CMTl1GwtlbK2DtIH6Dp1tBSxV9zxUOM/edit
We should use it.