Re:Re: can one keyspace has two groups of "initial_token"?
Thank you very much！ Could you please tell why the developers choose the design that token is a cluster setting but not a column family one？ At 2011-12-18 01:44:23,"Edward Capriolo" wrote: No. Token and partitioner are cluster wide settings. You have to run multiple instances of Cassandra. 2011/12/17 魏金仙 My question is whether two groups of "initial_token" can coexist since our goal is to partition data of each column family uniformly on 5 nodes. I deployed Cassandra 0.7.4 on a cluster of 5 nodes, using orderPreservingPartitioner. Two column families named CF1 and CF2 are created on one keyspace. Keys of CF1 go as a0,a1,a2,a3... while those of CF2 are b0,b1,b2 Assume that the initial_tokens should be set as a0,a2,a4,a6,a8 if you want to achieve load balance of CF1 and as b0,b2,b4,b6,b8 for CF2. Is it possible two set "initial_token" as: initial_token: a0,b0? Thank you!
Re: Re: can one keyspace has two groups of "initial_token"?
It has been mentioned before. In a nutshell the answer is "Do not use ByteOrderPartitioner". It has the problem you mentioned multiple-column families with different distributions as well as hot-spot problems. Moving the token data to a column family level would be a large change. it only helps the BOP/OOP, which along with supercolumns are highly NOT suggested in all cases, but people still seem to gravitate to them. 2011/12/18 魏金仙 > Thank you very much！ > Could you please tell why the developers choose the design that token is a > cluster setting but not a column family one？ > > > At 2011-12-18 01:44:23,"Edward Capriolo" wrote: > > No. Token and partitioner are cluster wide settings. You have to run > multiple instances of Cassandra. > > 2011/12/17 魏金仙 > >> My question is whether two groups of "initial_token" can coexist since >> our goal is to partition data of each column family uniformly on 5 nodes. >> >> I deployed Cassandra 0.7.4 on a cluster of 5 nodes, using >> orderPreservingPartitioner. Two column families named CF1 and CF2 are >> created on one keyspace. Keys of CF1 go as a0,a1,a2,a3... while those of >> CF2 are b0,b1,b2 >> Assume that the initial_tokens should be set as a0,a2,a4,a6,a8 if you >> want to achieve load balance of CF1 and as b0,b2,b4,b6,b8 for CF2. >> Is it possible two set "initial_token" as: initial_token: a0,b0? >> >> Thank you! >> >> >> > > >
Re: memory estimate for each key in the key cache
to be blunt, this doesn't sound right to me, unless it's doing something rather more clever to manage the memory. i mocked up a simple class containing a byte[], ByteBuffer and long, and the shallow size alone is 32 bytes. deep size with a byte[16], 1-byte bytebuffer and long is 132. this is a on a 64-bit jvm on win x64, but is consistent(ish) with what i've seen in the past on linux jvms. the actual code has rather more objects than this (it's a map, it has a pair, decoratedKey) so would be quite a bit bigger per key. On 17/12/2011 03:42, Brandon Williams wrote: On Fri, Dec 16, 2011 at 9:31 PM, Dave Brosius wrote: Wow, Java is a lot better than I thought if it can perform that kind of magic. I'm guessing the wiki information is just old and out of date. It's probably more like 60 + sizeof(key) With jamm and MAT it's fairly easy to test. The number is accurate last I checked. -Brandon
Re: CPU bound workload
>> I would first eliminate or confirm any GC hypothesis by running all >> nodes with -XX:+PrintGC -XX:+PrintGCDetails -XX:+PrintGCTimeStamps >> -XX:+PrintGCDateStamps. > > Is full GC not being logged through GCInspector with the defaults ? The GC inspect tries its best, but it's polling. Unfortunately the JDK provides no way to properly monitor for GC events within the Java application. The GC inspector can miss a GC. Also, the GC inspector only tells you time + type of GC; a GC log will provide all sorts of details. -- / Peter Schuller (@scode, http://worldmodscode.wordpress.com)
