keith-turner commented on code in PR #5301:
URL: https://github.com/apache/accumulo/pull/5301#discussion_r1956735562


##########
core/src/main/java/org/apache/accumulo/core/conf/Property.java:
##########
@@ -429,15 +429,41 @@ public enum Property {
   
MANAGER_FATE_METRICS_MIN_UPDATE_INTERVAL("manager.fate.metrics.min.update.interval",
 "60s",
       PropertyType.TIMEDURATION, "Limit calls from metric sinks to zookeeper 
to update interval.",
       "1.9.3"),
-  MANAGER_FATE_THREADPOOL_SIZE("manager.fate.threadpool.size", "64", 
PropertyType.COUNT,
-      "The number of threads used to run fault-tolerant executions (FATE)."
-          + " These are primarily table operations like merge.",
+  @Deprecated(since = "4.0.0")
+  MANAGER_FATE_THREADPOOL_SIZE("manager.fate.threadpool.size", "64",
+      PropertyType.FATE_THREADPOOL_SIZE,
+      "Previously, the number of threads used to run fault-tolerant executions 
(FATE)."
+          + " This is no longer used in 4.0+. MANAGER_FATE_USER_CONFIG and"
+          + " MANAGER_FATE_META_CONFIG are the replacement and must be set 
instead.",
       "1.4.3"),
+  MANAGER_FATE_USER_CONFIG("manager.fate.user.config", "{"
+      + 
"\"TABLE_CREATE,TABLE_DELETE,TABLE_RENAME,TABLE_ONLINE,TABLE_OFFLINE,NAMESPACE_CREATE,NAMESPACE_DELETE,NAMESPACE_RENAME,TABLE_TABLET_AVAILABILITY,SHUTDOWN_TSERVER\":
 1,"
+      + "\"TABLE_BULK_IMPORT2\": 2,"
+      + "\"TABLE_COMPACT,TABLE_CANCEL_COMPACT,COMMIT_COMPACTION\": 4,"

Review Comment:
   COMMIT_COMPACTION will likely be the most frequently executed fate operation 
in a system.  A single TABLE_COMPACT operation can cause N COMMIT_COMPACTION 
operations to eventually run.  Also system compactions that occur automatically 
as data is ingested can cause COMMIT_COMPACTION ops to run.  Should probably 
move TABLE_COMPACT,TABLE_CANCEL_COMPACT elsewhere w/ the other table ops as 
they are much less frequent operations and are always user initiated.
   
   I would recommend three default groups.   One for COMMIT_COMPACTION, one for 
SYSTEM_SPLIT, and one for everythnig else (I think everything else is the user 
initiated operations).  This will give dedicated resrouces to the the two 
system initated operations and keep them from intefering w/ the user initiated 
ops.



-- 
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: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to