Bug 755852

Summary: Message throughput drop (in comparison to 0.10) in a cluster when TTL set to 0
Product: Red Hat Enterprise MRG Reporter: Leonid Zhaldybin <lzhaldyb>
Component: qpid-cppAssignee: messaging-bugs <messaging-bugs>
Status: NEW --- QA Contact: Messaging QE <messaging-qe-bugs>
Severity: high Docs Contact:
Priority: medium    
Version: 2.1CC: jross
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On:    
Bug Blocks: 794748    

Description Leonid Zhaldybin 2011-11-22 08:42:06 UTC 
Description of problem:

While verifying this: https://issues.apache.org/jira/browse/QPID-3280, I came across the following problem.
The throughput in a cluster for the messages with TTL set to 0 seems to be at least twice lower for the  0.12 qpid version, when compared to the 0.10 version. In my tests, I used two virtual machines connected to a cluster, one of them running RHEL6 i686 and another one running RHEL6 x86_64. Sending 800000 of messages to such a cluster composed of 0.10 brokers, takes about 40-45 seconds, whereas sending the same amount of messages to a 0.12 cluster takes up to 105-110 seconds.
Swapping was turned off prior to the testing.
The 0.10 qpid brokers were running on RHEL6.1, 0.12 on RHEL6.2.

Version-Release number of selected component (if applicable):
qpid-cpp-server-0.12-6.el6
qpid-cpp-server-cluster-0.12-6.el6

How reproducible:
always

Steps to Reproduce:
1. Configure a number of qpid brokers ver.0.10 to run in a cluster.
2. Send a number of messages with TTL set to 0 to this cluster, wait till sending process is over, mark the duration.
3. Repeat steps for qpid brokers ver.0.12.
4. Compare durations for these two different qpid versions.
  
Actual results:
Sending messages with TTL=0 to the cluster consisting of the qpid brokers ver.0.12 lasts 2 - 3 times longer than sending them to the cluster consisting of the brokers ver.0.10.

Expected results:
The throughput of 0.12 cluster should not be lower than one of 0.10 cluster.

Additional info: