Use ConcurrentHashMap to protect FS_CACHE

[hanneskaeufler]

Closes #474

*Issue #, if available: #474

Description of changes:

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[markjschreiber]

The checks are passing now and the code looks good. Do you want to remove the draft status?

[hanneskaeufler]

The checks are passing now and the code looks good. Do you want to remove the draft status?

Sure! To be transparent though: I have not verified whatsoever that this solves the problem I was seeing in the referenced issue. It was more just a hunch that crossed my mind and I wanted to get to paper. I see no particular risk with this either, so we might as well get it in? Not sure about test automation here ... :/

[markjschreiber]

The checks are passing now and the code looks good. Do you want to remove the draft status?

Sure! To be transparent though: I have not verified whatsoever that this solves the problem I was seeing in the referenced issue. It was more just a hunch that crossed my mind and I wanted to get to paper. I see no particular risk with this either, so we might as well get it in? Not sure about test automation here ... :/

Agreed, there is no risk. There might be a slight performance drop but should be trivial compared to the network IO we have to do with S3. I don't know that you could unit test this other than to ensure that the Map is an instance of ConcurrentMap just to make sure nobody changes it back again without good reason.

I assume you would want to test it with your parallel test framework and it is always a good idea to try running the apps in the examples dir to ensure nothing weird happens.