Running stresstest to see how healthy a webserver is !?

hello dear OpenSuse-Experts hello and good day,

**the issue and topic of today: **running stresstest to check how healthy the server is - and to find out why a script sometimes stopps working:

note: the server runs on OpenSuse

Operating system	SuSE Linux 12.3
Webmin version	1.910 	Authentic theme version	19.33 
Time on system	Wednesday, November 27, 2019 6:59 PM	Kernel and CPU	Linux 4.4.148-ps on x86_64

in the past few weeks i have had some issues on a server - with Wordpress and Limesurvey. I also have openend a thread here. Now the site is up again and runs well. The limesurvey-page often has thrown a page like ERR_EMPTY_RESPONSE Error (both on FireFox and on Chrome)

Since 24 h. the behaviour is better - the site is up again. see path: http://www.f-s-j.de/limesurvey/index.php/admin/index
limesurvey :: https://www.limesurvey.org/about-limesurvey/ version 3.20.02

it is installed in (between) a wordpress :: in other words: the limesurvey was installed under a folder inside my wordpress installation.
Some limesurvey-experts say that i will get all kind of issues over the time. And that it has to do with the installation-structure:

  • MyDomain.TLD/limesurvey/ (LimeSurvey)
  • MyDomain.TLD/ (Wordpress)

hypothesis: LimeSurvey and Wordpress tell the server to rewrite URLs for the browser. In my setup Wordpress just tries to overrule certain settings of LimeSurvey. The Limesurvey-folks recommend that i would need to check every change and update manually the htaccess file (since i am running apache).

final recommendation: To minimize work and trouble, i should install wordpress and limesurvey side by side.
we then need to redirect from MyDomain.TLD/ -> MyDomain.TLD/wordpress/, but then (after this step) i have clean folders and don’t cause delays and loops when it comes to URL accessing. This safe way of installation side by side includes a special procedure: to use /w/ for wordpress and /s/ for limesurvey will keep the URL short.

Again: the error way: ERR_EMPTY_RESPONSE - after it worked very well for weeks

what can i do now!?

btw. the wordpress on the server still runs well

**question: **can i perfom a so called stress- or perfomance-Test to see how healthy the server is!?
in other words: is there a Performance-, Last- und Stress-Test for a Webserver out there.

note: some data:
PHP Version 5.6.39
eg. wordpress-basic-installation
https://www.limesurvey.org/about-limesurvey/download
sometimes i get the error: ERR_EMPTY_RESPONSE Error

look forward to hear from you

regards

Hi
I would be more worried about a security test… that system is just ripe for the picking!

Have a look in the benchmark repository https://build.opensuse.org/project/show/benchmark

For example httperf.

hello dear Malcom good day - great to hear from you i am very very glad.

btw: the server will be replace very very soon. You are convincing me -with all you say. Just some ideas to share - bout the issue of **a emty_Response_error

**

well i also have discussed the issue with the limesurvey-user here:

after having mused bout the errors and the possible reasons - after having mused bout porting over the installation on subomains or a
side to side installation i suddenly saw that the site is coming back… :wink:

note: i guess that the errors ERR_EMPTY_RESPONSE - is rooted in a very very weak server - just that easy - is this possible - is this thinkable!?

Holch - one user of the limesurvey - he has got the very same installation - structure as i have.

the user holch: : in this thread you answer very interestingly :

I run Limesurvey and Wordpress in a very similar setting (Wordpress in root and LS in subfolder)
and I don’t have the same problems. So while I wouldn’t want to exclude this as the source of issues completely,
I would probably focus on other aspects first, as it works fine with me for quite some time now.
Of course, as Jelo says, any update in one of the two applications could start causing problems.

at the moment it goes very well - in other words i have no issues at the moment: Holch - is it thinkable that this error is rooted in a very slow server!? I have to admit that this seerver is very very old and will be replaced very soon (in a few weeks)

you have mentioned that you have installed limesurvey in (exactly) the same way as i did.
you also have mentioned that you did not have encountered any of the issues as i have encountered.

is this true!?

furthermore; i have googled this error - guess that it is this one:

Error 324 (ERR_EMPTY_RESPONSE): The server closed the connection without sending any data
php - Error 324 empty response on Application hosted by AWS - Stack Overflow

website. I don’t think it is Chrome browser bug either because when I open safari I don’t get the error, I get a white page.Not sure what is happening or why. Basically, I am looking for the root causes for this server error.
Any ideas would be much appreciated. If you need me to provide any specific info, let me know.
Come to think of it, this error started happening because the server is pretty slow!

cf: php - Error 324 empty response on Application hosted by AWS - Stack Overflow

  • Except a CURL shows me the HTML just fine, but it’s not loading up in a browser. What in the world could be causing something like this :frowning: Everything was fine until last night
  • @Till’s curl advice is brilliant :slight_smile:
  • what about to work with Session-stickyness:
    Edit: How do you manage this problem in the future when you need session stickiness? I have no idea. But there should be answers on this out there!
  • …In addition to this, GZIP needs to be disabled at the web server (apache, or other).

what do you say - is this thinkable that the server ist pretty slow!?
what bout the test-method of using curl - in the moment when encountering the issue- !?

the idea of Err Empty_Response_ being rooted in the slow server - well this would be great - since i change the server in the next few weeks. Is this idea of the possible roots thinkable - and convincing!?

note: as it runs very well at the moment i will take a closer look at the behaviour - and will try to find out if /(when) the error will be appear again.

dear Malcom - i look forward to hear from you
regards

Hi
Until you move to something that is current (php, the operating system etc) and supportable both from an openSUSE and upstream point of view, I don’t think you would get any useful help… :frowning:

good day dear Malcom,

many thanks for the reply. Well youre right - i need to do all that is mentioned by you.
i will talk to the serveradmin

again many thanks for your support and all the help throughout the years

regards

Probably the first thing to note is that ERR_EMPTY_RESPONSE is not a standard response that indicates a server-side problem… most often HTTP 400 errors which generally refer to website problems and HTTP 500 errors which generally refer to server (not website) problems. Verify by also checking your Server logs because oftentimes errors to the client browser are masked to avoid embarrassment or hide possibly sensitive information from malicious hackers. In fact, by all accounts from an Internet search, the error is likely uniquely possible only when using a Chrome (Chrome family) web browser, and it means that your browser successfully established a network connection to the website, but the website didn’t respond.

You should be able to verify this by watching the bottom left corner of your Chrome web browser when you connect, it’s known as the Status Bar and as its name suggests should tell you what is going on in your web browser.

Especially if you experience problems only with one machine or one web browser,
You should immediately be able to eliminate any server-side problems as a likely cause and focus on why your web browser could be slow or faulty…
Some common things to do to resolve client browser-side problems…

  • Clear your cache and maybe your cookies as well
  • Eliminate or try to set your cache as tiny as possible, especially if you use your machine for support or development.
  • Often today, web browsers won’t allow you to zero the cache size completely… Therefor, always remember to “reload” the web page to ensure your web browser isn’t trying to read from your local browser cache.
  • Of course, there can be a networking problem but consider this possibility last, nowadays I’d think that highly capable broadband connections should be standard and unless someone is under <very> heavy load, like hundreds of simultaneous connections minimally I’d be surprised that common hardware and a default configuration would cause this kind of error(but don’t overlook the possibility someone or something that’s not your server really could be consuming excessive resources)…

And, especially in this case of course…
Use some other web browser than Chrome/Chromium/Canary/etc.

The above are things anybody can do,
Of course anyone who is deeper into Website development can also launch the browser Developer tools to further analyze whatever might be a cause…

I’ll leave your question about load testing (aka your stress testing) for another time when it might be more likely relevant to your problem…

TSU