Github user suyogmapara commented on the issue:
https://github.com/apache/zookeeper/pull/567
Thanks @anmolnar for the review. Regarding randomization, I think they did
it for snapCount because snapshot is potentially expensive IO and entire
ensemble doing it at the same time may be an issue. This feature doesn't do
snapshot when we roll the log and just switches the file we are writing to. So
it may be okay if everyone do it at the same time. Please let me know if you
disagree.---
