Comment 3 for bug 1072538

I think we just should modify the init script to:

1. Wait for the server to become available forever (rather than just for the hard-coded timeout of 14 seconds)
2. If startup takes longer than a predefine threshold, print a warning, something along the lines "Waited for more than N seconds, it might be because buffer pool restore is in progress and innodb_blocking_buffer_pool_restore is enabled".

I don't see a more elegant solution. It's a conflict of 2 correctly working features.