Possible concurrent issue with deployerMap in DeploymentEngine

[Kishanthan Thangarajah <ks...@gmail.com>]

Hi Devs,

There is a possible issue which can arise with the deployerMap (HashMap)
object in deployment engine. The deplyerMap object is accessed in two
places.
1. When adding a new deployer.
2. During the scheduler task run, which in turn calls
RepositoryListener.loadOtherDirectories.

If the add deployer method is called while the scheduler task is running,
then that can lead to ConcurrentModificationException. This is already
reported in https://issues.apache.org/jira/browse/AXIS2-5152. The fix
suggested in there is to use ConcurretnHashMap. But still the issue can
arise, since there is another second level map (which holds the deployer -
extension) within this. This still can cause the same concurrent
modification exception.

We can fix this by synchronizing the complete deployerMap object in those
places that is is accessed. But the concern I have is, will this cause any
impact on performance?

Thoughts?

Thanks,
Kishanthan.