Synchronize RefreshIsCurrent


It's possible to trigger RefreshIsCurrent from many threads, which causes contention on the disk.

We should prevent re-entry into this method.


Zooba wrote Jun 3, 2014 at 10:14 PM

Two aspects to this issue:
  • Multiple calls to the same instance on different threads will block.
    • These should assume the result is the same and not re-scan the disk.
  • Multiple calls to different instances will fight for IO time
    • Perhaps we should only allow x (4?) parallel calls to this function per-process