Github user mengxr commented on the pull request: https://github.com/apache/spark/pull/916#issuecomment-45992111 @colorant Tried the following with the new implementation: ~~~ val rdd = sc.parallelize(0 until 1000000000, 10000).flatMap(i => Iterator.fill(10)(0)) // 10^10 rdd.takeSample(false, 1).size rdd.takeSample(true, 1).size ~~~ Both worked well. We might need a better RNG for even smaller sampling probabilities. Another solution is set a lower bound in `comptueFractionForSampleSize`, e.g, `10^{-9}`. I prefer the latter to avoid using expensive RNGs. Could you run some tests and derive a good lower bound? Thanks!
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. ---