Details
-
Bug
-
Status: Closed
-
High
-
Resolution: Fixed
-
CernVM-FS 2.6
-
None
-
ANY
-
Description
I happened to notice that the lock implementation in the geo api is incorrect and completely ineffective; a new lock object is allocated every time, instead of sharing it between threads. I am amazed that it passes the parallelism test in 614-geoservice.