fapifta commented on issue #668: URL: https://github.com/apache/hadoop-ozone/pull/668#issuecomment-616631281
Hi @timmylicheng, as we discussed with @sodonnel I have taken a look at the PR, and we discussed about that sorting in the algorithm which feels to limit the performance of this algorithm. At this stage, as we do not have really large cluster implementations, we do not have a real measurable impact with the current solution, in the long run we might need to optimise the code, if we change the pipeline lifecycle, or if we have some really large clusters running into trouble, because of the performance of this code, but in that case we should certainly measure which part is the bottleneck, as at this point we can not be sure that the sort in this piece is the problem. (I feel that the way how we get the initial healthy nodes list is a similarly heavy part for example, though as I said at this point I am unsure.) So all in all I am +1 (non-binding) to commit this change, but we should note somewhere that the pipeline creation logic in the rack aware setup can cause startup performance problems as the algorithm is complex. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
