使用pycassa将键和行缓存设置为列族？

Question

我想知道 pycassa 是否可以将 keys_cached 和 rows_cached 字段设置为特定的列族（甚至是键空间）？如此处所示，使用 cassandra cli：http://www.datastax.com/docs/1.0/configuration/storage_configuration

我在文档中找到了采用key_cache_size 参数的alter_column_family 函数：http://pycassa.github.io/pycassa/api/pycassa/system_manager.html

但是当我在设置列族的键缓存大小（http://pycassa.github.io/pycassa/assorted/pycassa_shell.html）后使用 pycassaShell 检查缓存值时，它仍然告诉我：

Row Cache:                       None%
Key Cache:                       None%

而且两者之后的性能/内存使用也没有差异，并且由于alter_column_family 采用**kwargs 参数，并且显然不检查它们，所以使用任何参数名称调用成功......所以我认为key_cache_size什么都没有……而且我没有找到关于什么是可能的可选参数的文档。

这是收到alter_column_family 呼叫时的 cassandra 日志：

INFO 17:39:26,338 Update ColumnFamily '53c7deadcc9b10271a2df9f0/B' From org.apache.cassandra.config.CFMetaData@3fbd01a[cfId=2dd38542-82ea-381f-be51-44a30af61f24,ksName=53c7deadcc9b10271a2df9f0,cfName=B,cfType=Standard,comparator=org.apache.cassandra.db.marshal.IntegerType,comment=,readRepairChance=0.1,dclocalReadRepairChance=0.0,replicateOnWrite=true,gcGraceSeconds=864000,defaultValidator=org.apache.cassandra.db.marshal.DoubleType,keyValidator=org.apache.cassandra.db.marshal.IntegerType,minCompactionThreshold=4,maxCompactionThreshold=32,column_metadata={java.nio.HeapByteBuffer[pos=0 lim=3 cap=3]=ColumnDefinition{name=6b6579, validator=org.apache.cassandra.db.marshal.IntegerType, type=PARTITION_KEY, componentIndex=null, indexName=null, indexType=null}, java.nio.HeapByteBuffer[pos=0 lim=5 cap=5]=ColumnDefinition{name=76616c7565, validator=org.apache.cassandra.db.marshal.DoubleType, type=COMPACT_VALUE, componentIndex=null, indexName=null, indexType=null}, java.nio.HeapByteBuffer[pos=0 lim=7 cap=7]=ColumnDefinition{name=636f6c756d6e31, validator=org.apache.cassandra.db.marshal.IntegerType, type=CLUSTERING_KEY, componentIndex=null, indexName=null, indexType=null}},compactionStrategyClass=class org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy,compactionStrategyOptions={},compressionOptions={sstable_compression=org.apache.cassandra.io.compress.LZ4Compressor},bloomFilterFpChance=<null>,memtable_flush_period_in_ms=0,caching=KEYS_ONLY,defaultTimeToLive=0,speculative_retry=NONE,indexInterval=128,populateIoCacheOnFlush=false,droppedColumns={},triggers={}] To org.apache.cassandra.config.CFMetaData@11131f6f[cfId=2dd38542-82ea-381f-be51-44a30af61f24,ksName=53c7deadcc9b10271a2df9f0,cfName=B,cfType=Standard,comparator=org.apache.cassandra.db.marshal.IntegerType,comment=,readRepairChance=0.1,dclocalReadRepairChance=0.0,replicateOnWrite=true,gcGraceSeconds=864000,defaultValidator=org.apache.cassandra.db.marshal.DoubleType,keyValidator=org.apache.cassandra.db.marshal.IntegerType,minCompactionThreshold=4,maxCompactionThreshold=32,column_metadata={java.nio.HeapByteBuffer[pos=0 lim=7 cap=7]=ColumnDefinition{name=636f6c756d6e31, validator=org.apache.cassandra.db.marshal.IntegerType, type=CLUSTERING_KEY, componentIndex=null, indexName=null, indexType=null}, java.nio.HeapByteBuffer[pos=0 lim=3 cap=3]=ColumnDefinition{name=6b6579, validator=org.apache.cassandra.db.marshal.IntegerType, type=PARTITION_KEY, componentIndex=null, indexName=null, indexType=null}, java.nio.HeapByteBuffer[pos=0 lim=5 cap=5]=ColumnDefinition{name=76616c7565, validator=org.apache.cassandra.db.marshal.DoubleType, type=COMPACT_VALUE, componentIndex=null, indexName=null, indexType=null}},compactionStrategyClass=class org.apache.cassandra.db.compaction.SizeTieredCompactionStrategy,compactionStrategyOptions={},compressionOptions={sstable_compression=org.apache.cassandra.io.compress.LZ4Compressor},bloomFilterFpChance=<null>,memtable_flush_period_in_ms=0,caching=KEYS_ONLY,defaultTimeToLive=0,speculative_retry=NONE,indexInterval=128,populateIoCacheOnFlush=false,droppedColumns={},triggers={}]
INFO 17:39:26,349 CFS(Keyspace='system', ColumnFamily='schema_columnfamilies') liveRatio is 5.344978165938865 (just-counted was 5.344978165938865).  calculation took 0ms for 25 cells
INFO 17:39:26,349 Enqueuing flush of Memtable-schema_keyspaces@2042821730(138/8832 serialized/live bytes, 3 ops)
WARN 17:39:26,350 setting live ratio to maximum of 64.0 instead of Infinity
INFO 17:39:26,351 CFS(Keyspace='system', ColumnFamily='schema_keyspaces') liveRatio is 64.0 (just-counted was 64.0).  calculation took 1ms for 0 cells
INFO 17:39:26,351 Writing Memtable-schema_keyspaces@2042821730(138/8832 serialized/live bytes, 3 ops)
INFO 17:39:26,368 Completed flushing var/lib/cassandra/data/system/schema_keyspaces/system-schema_keyspaces-jb-79-Data.db (177 bytes) for commitlog position ReplayPosition(segmentId=1405696976943, position=238330)
INFO 17:39:26,373 Enqueuing flush of Memtable-schema_columnfamilies@200691258(1145/6120 serialized/live bytes, 25 ops)
INFO 17:39:26,373 Writing Memtable-schema_columnfamilies@200691258(1145/6120 serialized/live bytes, 25 ops)
INFO 17:39:26,381 Completed flushing var/lib/cassandra/data/system/schema_columnfamilies/system-schema_columnfamilies-jb-74-Data.db (787 bytes) for commitlog position ReplayPosition(segmentId=1405696976943, position=238330)

有一个有趣的caching=KEYS_ONLY 参数，它似乎与 pycassaShell 输出相矛盾，但与此键缓存大小的大小无关，与行缓存无关。无论我输入key_cache_size=200000、0 还是None，我都会得到相同的输出。

那么有人知道如何通过 pycassa 做到这一点吗？

Answer 1

您使用的是哪个版本的 Cassandra？ pycassa 在列族属性方面有点过时了（尤其是在 pycassaShell 中）。

key_cache_size 和 row_cache_size 在 Cassandra 1.1 的 cassandra.yaml 中被替换为全局选项（key_cache_size_in_mb、row_cache_size_in_mb）。 caching 选项是您应该为每个列族设置的选项，它可以设置为 KEYS_ONLY、ROWS_ONLY、ALL 或 NONE。

附带说明一下，此时您通常应该使用 cqlsh 而不是 cassandra-cli 和 python CQL driver 而不是 pycassa（我维护两者），尤其是对于新项目。