Chapter 4. Basic configuration options of Shenandoah garbage collector

Shenandoah garbage collector (GC) has the following basic configuration options:

Print the individual GC timing.
Print the heuristics decisions, which might shed light on outliers, if any.

Print the summary table on Shenandoah internal timings at the end of the run.

It is best to run this with logging enabled. This summary table conveys important information about GC performance. Heuristics logs are useful to figure out GC outliers.

Commit heap pages into memory and helps to reduce latency hiccups.
-Xms and -Xmx
Making the heap non-resizeable with -Xms = -Xmx reduces difficulties with heap management. Along with AlwaysPreTouch, the -Xms = -Xmx commit all memory on startup, which avoids difficulties when memory is finally used. -Xms also defines the low boundary for memory uncommit, so with -Xms = -Xmx all memory stays committed. If you want to configure Shenandoah for a lower footprint, then setting lower -Xms is recommended. You need to decide how low to set it to balance the commit/uncommit overhead versus memory footprint. In many cases, you can set -Xms arbitrarily low.
Enables hugetlbfs Linux support.
Enables huge pages transparently. With transparent huge pages, it is recommended to set /sys/kernel/mm/transparent_hugepage/enabled and /sys/kernel/mm/transparent_hugepage/defrag to madvise. When running with AlwaysPreTouch, it will also pay the defrag tool costs upfront at startup.
While Shenandoah does not support NUMA explicitly yet, it is a good idea to enable NUMA interleaving on multi-socket hosts. Coupled with AlwaysPreTouch, it provides better performance than the default out-of-the-box configuration.
There is a tradeoff between uncontended (biased) locking throughput, and the safepoints JVM does to enable and disable them. For latency-oriented workloads, turn biased locking off.
Invoking System.gc() from user code forces Shenandoah to perform additional GC cycle. It usually does not harm, as -XX:+ExplicitGCInvokesConcurrent gets enabled by default, which means the concurrent GC cycle would be invoked, not the STW Full GC.

Revised on 2023-11-14 21:13:41 UTC