----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/51696/#review148270 -----------------------------------------------------------
Ship it! Ship It! - Udo Kohlmeyer On Sept. 7, 2016, 4:25 p.m., Bruce Schuchardt wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/51696/ > ----------------------------------------------------------- > > (Updated Sept. 7, 2016, 4:25 p.m.) > > > Review request for geode, Anthony Baker, Hitesh Khamesra, Udo Kohlmeyer, and > Dan Smith. > > > Repository: geode > > > Description > ------- > > Looking through our distributedTest output I see a large number of full GCs > initiated by Metaspace cleanup. Some of these GCs are taking as long as 70 > seconds and cause nodes to be kicked out of the distributed system. For Java > 8 it's recommended that you estimate requirements for Metaspace and set > XXMetaspaceSize to avoid these full GCs. Our REST tests seem to need nearly > half a gigabyte of metaspace so I've set it accordingly. An AWS precheckin > run with this change showed only one GC initiated by metaspace management, in > the geode-web distributedTest suite lasting 1/2 second. > > > Diffs > ----- > > > geode-core/src/test/java/com/gemstone/gemfire/test/dunit/standalone/ProcessManager.java > f938c1a88cfec459efe4c87b8be41a6fa5c291e4 > > Diff: https://reviews.apache.org/r/51696/diff/ > > > Testing > ------- > > precheckin > > > Thanks, > > Bruce Schuchardt > >