I have not seen the issue after changing the commit log segment size to 1024MB. 


tpstats output:
Pool Name                    Active   Pending      Completed   Blocked  All 
time blocked
ReadStage                         0         0              0         0          
       0
RequestResponseStage              0         0              0         0          
       0
MutationStage                    32        40        2526143         0          
       0
ReadRepairStage                   0         0              0         0          
       0
ReplicateOnWriteStage             0         0              0         0          
       0
GossipStage                       0         0              0         0          
       0
AntiEntropyStage                  0         0              0         0          
       0
MigrationStage                    0         0              3         0          
       0
MemoryMeter                       0         0          24752         0          
       0
MemtablePostFlusher               1        19          12939         0          
       0
FlushWriter                       6        10          12442         1          
    2940
MiscStage                         0         0              0         0          
       0
PendingRangeCalculator            0         0              1         0          
       0
commitlog_archiver                0         0              0         0          
       0
InternalResponseStage             0         0              0         0          
       0
HintedHandoff                     0         0              0         0          
       0

Message type           Dropped
RANGE_SLICE                  0
READ_REPAIR                  0
PAGED_RANGE                  0
BINARY                       0
READ                         0
MUTATION                     0
_TRACE                       0
REQUEST_RESPONSE             0
COUNTER_MUTATION             0




On Saturday, 19 July 2014 1:32 AM, Robert Coli <rc...@eventbrite.com> wrote:
 


On Mon, Jul 7, 2014 at 9:30 PM, Bhaskar Singhal <bhaskarsing...@yahoo.com> 
wrote:

I am using Cassandra 2.0.7 (with default settings and 16GB heap on quad core 
ubuntu server with 32gb ram)

16GB of heap will lead to significant GC pauses, and probably will not improve 
total performance versus 8gb heap.
 
I continue to maintain that your "problem" is that you are writing faster than 
you can flush.

Paste the output of "nodetool tpstats"?

=Rob

Reply via email to