Details
-
New Feature
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
-
Unknown
Description
The file consumer has a number of read locks that are file based, such as marker files, java FileLock api etc.
But they are not guaranteed to work in a clustered environment where you want multiple nodes with Camel file routes compete for the same files.
The file consumer do have both an in progress and idempontent repository associated with it, but they come after the read lock. And they are implemented assuming that the read lock would be atomic and if the lock was granted its the sole reader of the file. So we need to strength the weakest link which was the read lock.
But using a read lock that reuse the existing idempotent implementations we can tap into clustering with zookeeper - hazelcast etc. or the shared JDBC database etc.