My Situation
I am using the SSH Command tool to retrieve log info from a remote Linux server using the tail command. A separate thread group logs in users which causes the log file to update with the login information. To get the data I need, I use a regex extractor on the response data. For every thread jmeter creates a new ssh connection to retrieve the new messages in that log file.
Here's my current setup:
Thread Group
- Get random user
- Login User
- SSH into server using tail command
- Extract relevant data
My Question
Instead of sshing into the Linux server every thread, causing extra load and log messages, I want to connect once at the start of the test and continuously extract from the log file using the tail -f command, for example, combined with the regex. Is this possible?
I would say that it is not possible or at least not easy.
In order to be able to extract the data from the response using Regular Expression Extractor you need to have SampleResult, it means that the SSH Command request must be completed. If you're going to use tail -f the request will never end.
If you want to minimize the number of connections you can consider using JSch library to establish the connection once and execute the commands within the bounds of a single Session.
If the above solution makes sense - consider migrating to JSR223 Sampler and Groovy language, you can find example code to execute remote command over SSH in i.e. Exec.java class
Related
I am using jmeter to test at my dev server.
The Scenario is like This.
0. Turn OFF all firewalls both local PC(so called, HOST) and client PC(so called, CLIENT)
Turn on Jmeter at my HOST
--> add Thread Group, bzm-Parellel Controller. I am not certain at this point
Connect to CLIENT (once)
-> maybe, by SSH Command or REMOTE Start
execute my test script at CLIENT (several times, more than 100 times)
-> such as, 'ls' 'pwd' 'mkdir dir123' 'ls' IN A ROW!!
-> maybe, by OS Process Sampler. I am not certain at this point
get result of (3) at my HOST jmeter by View Results Tree
This is the scenario that I Thought
Can anyone help me with this issue.
Cuz, there's too many Samplers and less information, I'm suffering such a tough moments.
Thank you for reading.
Turning off firewall is not the best idea, just open the port you will be using, default port for SSH is 22 and normally it's getting opened more or less automatically when you install OpenSSH server
I don't think you need Parallel Controller, it has very specific use cases like simulating AJAX requests, it will be sufficient to specify the desired number of users/loops/test duration on Thread Group level
Remote start is for JMeter distributed testing, if you want to run a shell command on the client use OS Process Sampler or SSH Command sampler, see How to Run External Commands and Programs Locally and Remotely from JMeter article for more details.
The same is for point 2, if you need to create a directory depending on your HOST and CLIENT operating systems you need to choose one of the aforementioned samplers. Just be aware that only first operation/iteration succeeds, all the subsequent attempts you will get cannot create directory ‘dir123’: File exists. I'm also not certain what you're trying to test here? SSH server performance? Operating system performance? Network performance?
If you add View Results Tree listener to your Test Plan and run your test in GUI mode it will automatically catch all the Samplers results
Below is the question, I was asked in an interview,
Datacenter has 10000 servers.We have a single syslog driver which collates all the logs from all the servers in the datacenter and writes it in a single file called syslog.log
Let's say the datacenter has 1000 Admins.At any point of time any admin can login to syslog server and invoke a command say
getlog --serverid --severity
And the command should continuously tail the logs matching the conditions provided by the user untill he interupts.
Any number of users can concurrently login to this server and run this command. His request should be honoured, but with one condition, at any given point in time there can be only one file descriptor open for syslog.log file.
Implement getlog such that it satisfies the above condition.
I told my approach as Critical Section problem, we can use mutex/semaphore to lock the file until a user finishes. But he is expecting something like Client-Server Model.
How to serve this functionality using client and server architecture?
What is the best approach to solve this?
I am trying to do a LDAP mod operation through Jmeter. Expected behavior - Jmeter would hit server A which in turn would hit server B. Actual modification operation would happen at server B. Server B would complete the operation and give response to server A which in turn would respond to Jmeter.
Now the issue is, Jmeter is always getting the "Referral" response message. However, manually we are able to change the password after hitting server A from different remote server.
Could you someone please suggest how to overcome this?
I am assuming this has been resolved. Just in case you are still wondering, #Rohan , my understanding is that you run jmeter on the command line:
$ jmeter -Jjava.naming.referral=true -n -t testplan.jmx -l log.jtl
JMeter won't have specific behaviour of its own. You will need to tell it to follow referrals by setting java.naming.referral property appropriately in the jndi.properties mechanism defined in the documentation for the JNDI LDAP provider, which you should already have in place for your application if you expect it to behave that way.
I have connected remotely to a QMgr via MQ Explorer on Windows. The MQ server version is 7.5.0.1. I can put messages in SYSTEM.ADMIN.COMMAND.EVENT from MQ Explorer successfully and when I dump SYSTEM.ADMIN.COMMAND.EVENT, I can see my messages. As long as I know, I should be able to run PCF commands and MQSC commands via this channel. So, I put DISPLAY QMGR ALL message inside this queue and I can successfully see this message on MQ Server. My question is how can I run this message remotely via this channel? Thanks.
IBM Doc indicates that I should be able to receive the command result in SYSTEM.MQSC.REPLY.QUEUE. But I can not browse this queue from client MQ Explorer. The queue type for this queue is Model.
Couple of problems here.
First, you are using the wrong queue. The command server listens on SYSTEM.ADMIN.COMMAND.QUEUE. The queue to which you are sending messages, SYSTEM.ADMIN.COMMAND.EVENT is the queue to which the QMgr puts event messages after executing commands, provided of course that command events are enabled.
The second problem, as Jason mentions, is that the runmqsc processor takes human-readable script and converts it into commands the QMgr can understand. Passing textual commands directly to the command server won't work.
Typically we do what you want by passing the commands to runmqsc directly such as...
echo DISPLAY QMGR ALL | runmqsc MYQMGRNAME
If you require the ability to do this as a client, then you want to either download SupportPac MO72, or head over to MQ Gem and pick up a copy of MQSCX. Either of these will accept the command above on a local queue manager, and both can also be supplied with MQ Channel params and connect to a remote QMgr.
In addition to this basic functionality, the MQSCX product also has its own internal script parsing and execution. Suppose, for instance, that you want to do something depending on the command level of the QMgr.
Using runmqsc you could issue the command above, filter the resulting 2-column output through grep, awk, or similar, then capture the final output into a variable. You might need to do this multiple times to capture multiple values, invoking a new runmqsc each time and parsing the output in your script. You must then generate the string for the actual command you wanted to run when you started all this, and pass it to another invocation of runmqsc.
Alternatively, MQSCX lets you issue the DISPLAY command, then reference the resulting values directly by name. For example, you can pass MQSCX a couple lines of script telling it to inquire on the QMgr and then take a conditional action based on the command version, all without ever having to drop back into shell, bat or Perl script.
Full disclosure, I do not work for or get a commission from MQ Gem. I just don't like to beat my head against the wall writing 100 lines of code where 2 will do. If you do any amount of MQSC scripting, the ROI on MQSCX is measured in minutes. And it happens to be 100% on-topic as an answer to this question.
The command server doesnt process textual messages, it processes PCF messages. You need to build a message in PCF format and it can be processed. See http://www-01.ibm.com/support/knowledgecenter/SSFKSJ_8.0.0/com.ibm.mq.adm.doc/q019980_.htm
Ideally you would use real PCF format but there is a PCF format where you can send MQSC commands ('escaped' PCF - see here http://www-01.ibm.com/support/knowledgecenter/SSFKSJ_8.0.0/com.ibm.mq.ref.adm.doc/q087230_.htm?lang=en)
I am currently using the Trilead ssh2 library and when i try to execute multiple commands (using execCommand) in the same session, I get "a remote execution has already started" error.
Just wanted clarify, the session is limited to one system command; does that mean I can only send one command through execCommand() per connection? Is there any other alternative besides injecting multiple commands with semicolons?