https://bz.apache.org/bugzilla/show_bug.cgi?id=62795
Bug ID: 62795
Summary: JMeter master sometimes ends distributed test even
though some of the slaves have not finished
Product: JMeter
Version: 4.0
Hardware: PC
Status: NEW
Severity: normal
Priority: P2
Component: Main
Assignee: [email protected]
Reporter: [email protected]
Target Milestone: JMETER_5.1
Created attachment 36185
--> https://bz.apache.org/bugzilla/attachment.cgi?id=36185&action=edit
Archive with logs from master and slaves
- Description: JMeter master sometimes ends distributed test even though some
of the slaves have not finished
- Environment:
-- 1 Master: ubuntu 16.04, JMeter version: 4
-- 2 Slaves: ubuntu 16.04, JMeter version: 4
All machines are running: openjdk version "1.8.0_181"
The .jmx file is checking if a variable from user.properties is set to true,
and if it is, it then runs some code.
Steps:
1.
./jmeter -n -t /fake-path/prepare-environment.jmx -r
-Djava.rmi.server.hostname=MASTERIP
Results:
In some conditions, the master decides that the test has ended even thought one
of the slaves has not finished.
It seems this is happening when one of the slaves finishes before the other
slaves starts.
As seen in jmeter.log the second slave finishes "Finished remote host:
10.17.41.203" and then the first slaves starts "Started remote host:
10.17.41.202". After this the master creates the dashboard while one of the
slaves is still running.
If i change the logic of the two slaves and jmeter starts the remote engines in
the same order the test fill finish successful, but i can not rely on this.
As a workaround i put both slaves to sleep for a couple of seconds at the start
of test so one does not finish before the other starts.
I do not really know how jmeter decides which slave will start first.
Attached jmeter.log for master and both slaves.
--
You are receiving this mail because:
You are the assignee for the bug.
