update default commitlog segment and total size

patch by jbellis; reviewed by vijay for CASSANDRA-4422

CHANGES.txt

@@ -1,4 +1,6 @@
 1.1.3
+ * update default commitlog segment size to 32MB and total commitlog
+   size to 32/1024 MB for 32/64 bit JVMs, respectively (CASSANDRA-4422)
  * avoid using global partitioner to estimate ranges in index sstables
    (CASSANDRA-4403)
  * restore pre-CASSANDRA-3862 approach to removing expired tombstones

conf/cassandra.yaml

@@ -155,13 +155,16 @@ saved_caches_directory: /var/lib/cassandra/saved_caches
 commitlog_sync: periodic
 commitlog_sync_period_in_ms: 10000
 
-# Configure  the Size of the individual Commitlog file. The 
-# default is 128 MB, which is almost always fine, but if you are
-# archiving commitlog segments (see commitlog_archiving.properties),
-# then you probably want a finer granularity of archiving; 16 MB
-# is reasonable.
+# The size of the individual commitlog file segments.  A commitlog
+# segment may be archived, deleted, or recycled once all the data
+# in it (potentally from each columnfamily in the system) has been 
+# flushed to sstables.  
 #
-# commitlog_segment_size_in_mb: 128
+# The default size is 32, which is almost always fine, but if you are
+# archiving commitlog segments (see commitlog_archiving.properties),
+# then you probably want a finer granularity of archiving; 8 or 16 MB
+# is reasonable.
+commitlog_segment_size_in_mb: 32
 
 # any class that implements the SeedProvider interface and has a
 # constructor that takes a Map<String, String> of parameters will do.
@@ -218,10 +221,14 @@ concurrent_writes: 32
 # If omitted, Cassandra will set it to 1/3 of the heap.
 # memtable_total_space_in_mb: 2048
 
-# Total space to use for commitlogs. 
+# Total space to use for commitlogs.  Since commitlog segments are
+# mmapped, and hence use up address space, the default size is 32
+# on 32-bit JVMs, and 1024 on 64-bit JVMs.
+#
 # If space gets above this value (it will round up to the next nearest
 # segment multiple), Cassandra will flush every dirty CF in the oldest
-# segment and remove it.
+# segment and remove it.  So a small total commitlog space will tend
+# to cause more flush activity on less-active columnfamilies.
 # commitlog_total_space_in_mb: 4096
 
 # This sets the amount of memtable flush writer threads.  These will

src/java/org/apache/cassandra/config/Config.java

@@ -97,11 +97,11 @@ public class Config
 
     // Commit Log
     public String commitlog_directory;
-    public Integer commitlog_total_space_in_mb = 4096;
+    public Integer commitlog_total_space_in_mb;
     public CommitLogSync commitlog_sync;
     public Double commitlog_sync_batch_window_in_ms;
     public Integer commitlog_sync_period_in_ms;
-    public int commitlog_segment_size_in_mb = 128;
+    public int commitlog_segment_size_in_mb = 32;
 
     public String endpoint_snitch;
     public Boolean dynamic_snitch = true;

src/java/org/apache/cassandra/config/DatabaseDescriptor.java

@@ -142,6 +142,9 @@ public class DatabaseDescriptor
             Yaml yaml = new Yaml(new Loader(constructor));
             conf = (Config)yaml.load(input);
 
+            if (!System.getProperty("os.arch").contains("64"))
+                logger.info("32bit JVM detected.  It is recommended to run Cassandra on a 64bit JVM for better performance.");
+
             if (conf.commitlog_sync == null)
             {
                 throw new ConfigurationException("Missing required directive CommitLogSync");
@@ -172,6 +175,9 @@ public class DatabaseDescriptor
                 logger.debug("Syncing log with a period of " + conf.commitlog_sync_period_in_ms);
             }
 
+            if (conf.commitlog_total_space_in_mb == null)
+                conf.commitlog_total_space_in_mb = System.getProperty("os.arch").contains("64") ? 1024 : 32;
+
             /* evaluate the DiskAccessMode Config directive, which also affects indexAccessMode selection */
             if (conf.disk_access_mode == Config.DiskAccessMode.auto)
             {