rendezvous point
DESCRIPTION
rTRANSCRIPT
Rendezvous Point
Select the Priority transaction and give the Rendezvous point
Put conditions to the Rendezvous point
Release first radio button for (i.e if we have 50 users , when we give 100% - will wait in Rendez point till
50 users arrive in Rendez point if we give 40% then will wait for first 20 users, second 20 users , then
wait for the next 10 users (with condition of Timeout between Vusers).
For Second Radio Button
Release when some % of running Vusers running in Rendezvous .. (Expecting some % of running point in
the Run )
Like (Rendezvous if we are expecting to release 50 users at Rendezvous point we will able to see only 49
number in Rendezvous point)
50% case
Down Run Rende
50
37 8 5 releasing with 50% (i.e the users at rende is 6
and run users are 7 only , when ever user is moving from run to rendezvous (the run will be with 12 i.e
run 6 and rende 6 and rende 6 i.e released internally ( we are not able to vie w the same number)
24 8 5
11 8 5
5 3
2 2
3rd radio button is same as 1st button
We can disable the Vusers in Rendezvous point so that they can execute the other scenario
The Vusers will execute the other transaction with out waiting
Stress Test:
Application level stress Is nothing but removing Think time and pacing time.
Type of Test Executions:
1) Smoke Test: Validate the behavior of end to end test environment with minimum amount of
vusers.
Ramp Up: 50 Vusers with 5 scripts with ramp up with 10 secs for every user.
Duration: 30 mins
Ramp Down: Simultaneously.
Load Test: Test the behavior of the application with the expected load.
Ramp UP: 2 seconds
Duration : 1 hour (if ramp up is excluded then 45 mins is enough)
Ramp Down: Simultaneously.
Stress Test: Test the behavior of application by overloading the vusers load on the server.
Application level: will apply the application level stress by (ignoring ) removing (reducing) pacing and
think times.
Function level stress: Do the functional level test using rendezvous points
Vusers load: 500
Ramp Up: 2 secs
Duration : 1 hours (No pacing and No think time)
Pace time: As soon as the previous iteration ends.
Think time: Ignore think time
Ramp down: simultaneously.
Endurance Test: Test the behavior of an application in an prolonged duration.
Vusers Load: 500
Ramp Up: 15 secs
Duration: 12 hours
Ramp down: simultaneously.
Spike Test : Test the behavior of an application with dynamic work load characteristic changes at
runtime.
Increase or decrease the users use Run/Stop. We can increase the users with any load gererator.
Ramp UP: only for first batch of vusers
Duration: First 30 mins 200 users . 31-60 mins 400 vusers. 61-90 mins 500 vusers 90-150 mins total users
500 Next half an hour 300 users (decrement by 200) next 30 mins zero level.
Controller Basic concepts
Convert the scenario to the % Mode
when we give the total Vusers in Global Schedule the load will be calculated and taken from it.
Ramp Up is : Initializing the Vuser from load generator to Server.
If the ramp up is happening after Vuser Init then it is not satisfied with the concept because the users
already on The server already active.
Ramp up should be before Vuser Init
Always in the Initialize option (initialize each Vuser just before it runs.—Ready status )
Define SLA in LR directly
SL
SLA defined for AVg Transaction response so it is disabled
Any no of ranges can be defined.
We can define another SLA through check box
Unix shell scripting:
Unix Monitoring: Unix counters: Here there are no objects and instances only counters are available.
CPU utilization == % Processor time (windows)
Interrupts Rate = Interrupts /Sec (windows)
System Mode CPU utilization == Previleged time (Windows)
User Mode CPU utilization == User time(windows)
Page-in rate == Pages input/Sec
Page-out rate== Pages output/Sec
Paging rate == Pages/Sec
Swap in rate == Page reads/Sec
Swap out rate == Page writes/Sec
Average Load :
Disk Traffic :
Incoming packet rate
Outgoing packet rate
Communicating from a windows machine to Unix machine.
We need to have a Putty
Rm –r script name to remove the file or folder
When controller and server are sitting in different networks then we use Unix monitors / Perfmon
(windows)
Monitor the below sh (shell scripting files) . to get the unix or linux os level metrics
Iostat.sh , vmstat.sh, netstat.sh, topstat.sh, mpstat.sh, memstat.sh
To start a file
./startall.sh
WINSCP
First window (windows) 2nd window (unix)
Drag and drop any folder from one window to other window
interval= 10 Sun May 29 19:32:49 IST 2011
Linux
2.6.18-164.el5xen (LinuxServer) ########
avg-cpu: %user %nice %system %iowait %steal %idle
1.08 0.06 5.25 0.8 0.01 92.81
Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s avgrq-sz
avgqu-sz await svctm %util
sda 5.96 6.09 4.33 2.71 341.22 70.52 58.46 0.13 18.5 2.23 1.57
sda1 0.26 0 0.03 0 0.3 0 10.25 0 4.67 4.17 0.01
sda2 0 0 0 0 0 0 2 0 34 34 0.01
sda5 0.26 0 0.03 0 0.3 0 10.79 0 4.82 4 0.01
sda6 0.12 0 0.03 0 0.3 0 11.15 0 2.48 1.94 0.01
sda7 0.37 0 0.04 0 0.82 0 18.73 0 5.61 4.97 0.02
sda8 4.94 6.09 4.19 2.71 339.36 70.52 59.39 0.13 18.77 2.23 1.54
dm-0 0 0 9.01 8.82 339.01 70.52 22.97 0.65 36.48 0.86 1.54
dm-1 0 0 0.03 0 0.24 0 8 0 12.56 1.5 0
avg-cpu: %user %nice %system %iowait %steal %idle
15.03 0 84.97 0 0 0
Here avg-cpu line items should be copied into next sheet and then analyze 90%
avg-cpu: %user %nice %system %iowait %steal %idle
1.08 0.06 5.25 0.8 0.01 92.81
15.03 0 84.97 0 0 0
SQL Server Metrics:
SQL Server:
Buffer Manager: Buffer Cache Hit Ratio :
“It measures the % of buffer cache memory use in buffer manager (which is before the sql server and in
b/w the system and sql server)
SLA : Always the Cache Hit Ration should be >95%
Lazy writes/Sec: It shows the number of slow query writes/sec from buffer manager to database.
SLA : Always lazy writes/Sec should be zero.
Total Pages: == Reserved Pages + Stolen Pages + Free Pages + Used Pages.
Reserved Pages : the pages are ready for usage.
Free Pages: The pages which are already used and now freed.
Stolen Pages: The pages which are using for the miscellaneous purpose (Not for our application purpose)
Used Pages: The pages which are used under our application.
Note: Always the stolen pages count should be less than the used pages count.
SQL Server General Statistics
Logical Connections: It shows the number of the logical connections are established at database.
SLA : It is always depending on the number of Vusers load.
User Connections: It shows the number of the user/physical connections are established at database.
SLA : It is always depending on the number of Vusers load.
SQL Server Locks:
Number of Dead Locks/Sec:
It shows the number of lock requests/sec in the execution at database.
SLA : Always the number of dead locks/sec should be zero.
SQL Server: SQL Errors :
Errors/Sec:
DB Off line Errors: compilation errors,
“It returns the error messages which are belongs to query compilations”
SLA : Always the offline errors should be less like in 2 digits (10-40 etc)
Info Errors : Exceptions (divide by zero )
“It returns the information on errors occurs in the execution”
SLA: It is always depending on failed transaction rate.
Kill Connection Errors: Failed to connect
“ It shows the number of connections failed errors in the execution”
User Errors: Login fails.
“It shows the number of login and log out errors in the execution”
SLA : Number of Login/ Log out failed attempts in the execution.