>Should I have this enabled or disabled and for all or some of the requests?
Your browser normally sends Keep Alive, so you should probably enable it. However what you could check is whether disabling this for all requests makes the response time uniform(ly slow) , in which case you know why the first request takes more time. if it doesnt make much difference you'd have to see whether it is something on the server (commonly cache or session) >Is there a hardware requirement for JMeter?? No. But there is a point in time with increasing load, when your jmeter client itself cant handle the load you want to generate. There can be no definitive answer for this , monitor your machine and see what it can bear. regards deepak On Wed, Nov 18, 2009 at 5:49 AM, cmrz <[email protected]> wrote: > > I do have KeepAlive on all of the requests. Should I have this enabled or > disabled and for all or some of the requests? What does it do? I > searched > documentation but can't find a definitive answer. > > for 2nd question --- I will be investigating this further today / tomorrow. > and will update . Is there a hardware requirement for JMeter?? -- also > searched but can't find. > > Machines I am running on have 512 and 1gb ram and 2.2 celerons and 2.4 P4s > respectably. > > thanks!!! > > > Deepak Shetty wrote: > > > > 1) do you have KeepAlive on your requests? Does your application load > > anything into the session the very first time? > >>What I am seeing is that the very first call ( to bring up the login page > > )takes 3-4 seconds. The rest of the calls have a normal response time. > > When you say rest of the calls , do you mean the subsequent calls or the > > same login calls (in different iterations). Its quite possible that a > page > > has different load times from the rest (you might only have a problem if > > the > > same page shows vastly differing load times). Jmeter does not cache > > anything unless you added the Cache Manager to your test plan > > > >>2) I have a test that runs a report. -- 10 threads response times are > >>good. 25 threads -- response times are tripple! My question is, at > what > >>point does machine resources start to effect test results? > > The problem could be your client or your server. An easy way to check is > > run > > two separate Jmeter instances on two separate client machines with say > > half > > the threads on each (so your total load is still 25 threads) at the exact > > same time. if your response times are bad, then it means your server is > > overloaded. If they are good , it means that your client is overloaded > > (there are other ways to check this using network monitoring tools and > > other > > os specific perf tools, but the above is the easiest) > > > > On Tue, Nov 17, 2009 at 5:29 AM, cmrz <[email protected]> wrote: > > > >> > >> Hi, > >> > >> I have a couple of questions . . . > >> > >> 1) I am running Jmeter via command prompt. I am executing the login > >> script. > >> Test pulls up the login page, enters user name and password, clicks the > >> login button and logs the user in. > >> > >> What I am seeing is that the very first call ( to bring up the login > page > >> ) > >> takes 3-4 seconds. The rest of the calls have a normal response time. > >> When > >> I run the test for multiple threads, I see response time of 3-4 seconds > >> for > >> this very first initial call and the rest of the calls are ok. Thread 1 > >> - > >> first call 3-4 sec. Thread 2 first call - 0.2 sec. > >> > >> I did not enable caching - unless JMeter does it on its own? > >> > >> Why is this happening? > >> > >> > >> 2) I have a test that runs a report. -- 10 threads response times are > >> good. 25 threads -- response times are tripple! My question is, at > >> what > >> point does machine resources start to effect test results? > >> > >> thank you so much for help!! > >> -- > >> View this message in context: > >> http://old.nabble.com/response-time-anomolies-tp26389922p26389922.html > >> Sent from the JMeter - User mailing list archive at Nabble.com. > >> > >> > >> --------------------------------------------------------------------- > >> To unsubscribe, e-mail: [email protected] > >> For additional commands, e-mail: [email protected] > >> > >> > > > > > > -- > View this message in context: > http://old.nabble.com/response-time-anomolies-tp26389922p26408066.html > Sent from the JMeter - User mailing list archive at Nabble.com. > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
