kohlmu-pivotal commented on a change in pull request #7456:
URL: https://github.com/apache/geode/pull/7456#discussion_r832774433



##########
File path: 
geode-core/src/main/java/org/apache/geode/internal/cache/control/HeapMemoryMonitor.java
##########
@@ -144,82 +84,34 @@
   private final ResourceAdvisor resourceAdvisor;
   private final InternalCache cache;
   private final ResourceManagerStats stats;
-  private final TenuredHeapConsumptionMonitor tenuredHeapConsumptionMonitor;
 
-  @MutableForTesting
-  private static boolean testDisableMemoryUpdates = false;
   @MutableForTesting
   private static long testBytesUsedForThresholdSet = -1;
 
-  /**
-   * Determines if the name of the memory pool MXBean provided matches a list 
of known tenured pool
-   * names.
-   *
-   * Package private for testing.
-   * checkTenuredHeapConsumption
-   *
-   * @param memoryPoolMXBean The memory pool MXBean to check.
-   * @return True if the pool name matches a known tenured pool name, false 
otherwise.
-   */
-  static boolean isTenured(MemoryPoolMXBean memoryPoolMXBean) {
-    if (memoryPoolMXBean.getType() != MemoryType.HEAP) {
-      return false;
-    }
-
-    String name = memoryPoolMXBean.getName();
-
-    return name.equals("CMS Old Gen") // Sun Concurrent Mark Sweep GC
-        || name.equals("PS Old Gen") // Sun Parallel GC
-        || name.equals("G1 Old Gen") // Sun G1 GC
-        || name.equals("Old Space") // BEA JRockit 1.5, 1.6 GC
-        || name.equals("Tenured Gen") // Hitachi 1.5 GC
-        || name.equals("Java heap") // IBM 1.5, 1.6 GC
-        || name.equals("GenPauseless Old Gen") // azul C4/GPGC collector
-
-        // Allow an unknown pool name to monitor
-        || (HEAP_POOL != null && name.equals(HEAP_POOL));
-  }
+  @Immutable
+  private static final ServiceLoader<HeapUsageProvider> heapUsageMonitorLoader 
=
+      ServiceLoader.load(HeapUsageProvider.class);

Review comment:
       This approach seems quite limiting and not at all conducive to 
dependency injection. Also it would make it really hard to test, without having 
to make sure that the correct jars are on the classpath.
   
   Maybe the approach could be amended, where the constructor either has the 
HeapUsageProvider provided as an argument OR an identifier is passed in, which 
allows the constructor to find the suitable HeapUsageProvider (just in case 
there are more than 1 on the classpath).




-- 
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.

To unsubscribe, e-mail: notifications-unsubscr...@geode.apache.org

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Reply via email to