Cassandra - Executing actions with multiple threads

[Arjun]

Hi,

I get 0 throughput when I execute the actions with multiple threads(10,100). 

Data Store - Cassandra.

The below is the log for ViewProfileAction.

After init: Sun Mar 09 13:52:30 PDT 2014 Connected Starting benchmark. Maximum execution time specified as: 180 secs 0 sec: 0 actions; 0 sec: 0 operations; 10 sec: 5242 actions; 522.94 current acts/sec; 10 sec: 5242 operations; 522.94 current ops/sec; [PROFILE AverageResponseTime(us)=2438.39] 20 sec: 5242 actions; 0 current acts/sec; 20 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 30 sec: 5242 actions; 0 current acts/sec; 30 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 40 sec: 5242 actions; 0 current acts/sec; 40 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 50 sec: 5242 actions; 0 current acts/sec; 50 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 60 sec: 5242 actions; 0 current acts/sec; 60 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 70 sec: 5242 actions; 0 current acts/sec; 70 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 80 sec: 5242 actions; 0 current acts/sec; 80 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 90 sec: 5242 actions; 0 current acts/sec; 90 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 100 sec: 5242 actions; 0 current acts/sec; 100 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 110 sec: 5242 actions; 0 current acts/sec; 110 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 120 sec: 5242 actions; 0 current acts/sec; 120 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 130 sec: 5242 actions; 0 current acts/sec; 130 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 140 sec: 5242 actions; 0 current acts/sec; 140 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 150 sec: 5242 actions; 0 current acts/sec; 150 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 160 sec: 5242 actions; 0 current acts/sec; 160 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] 170 sec: 5242 actions; 0 current acts/sec; 170 sec: 5242 operations; 0 current ops/sec; [PROFILE AverageResponseTime(us)=?] Maximum time elapsed. Requesting stop for the workload.10 Stop request is set to :true Stop requested for workload. Now Joining!

[Sumita Barahmand]

Hi,

I'm not sure if I can tell what the problem is just by looking at the output provided. But it seems like it does some work in the 10 seconds and then it's not able to generate requests. So somehow your threads are getting blocked. 

Try putting a break point and figuring out in what line of your client the threads are getting blocked.

Sumita

--
You received this message because you are subscribed to the Google Groups "BG Social Benchmark" group.
To unsubscribe from this group and stop receiving emails from it, send an email to BG-SocialBenchm...@googlegroups.com.
Visit this group at http://groups.google.com/group/BG-SocialBenchmark.
For more options, visit https://groups.google.com/d/optout.

[Shahram Ghandeharizadeh]

Hi,
Examine the global variables you've declared to verify concurrent threads do not suffer from race conditions.  Pay special attention to "static" variables.

Simplify the workload, reducing it to one action only (say View Profile by setting its frequency to 1 and reducing the frequency of all other actions to zero) to see if your client executes with 10 and 100 threads.  Once this is works then add a new action (say List Friends).  Continue to do so until you identify the action that causes the observed throughput to zero.  Focus on this action by making setting its frequency 1 to see if it causes the throughput to drop down to zero by itself.  Debug this action.  If you need help, post the body of this action on this forum and we'll help.

Cheers,
Shahram.