You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Since a few week, the server i am running against is upping it's security, they enabled HTTPS, I have been asked to run my tests over HTTPS, to load this new 'feature'.
On Http, I have been able to load a 1000Clients at a Client per sec rate without any problem.
Once I enable Https, for the few first clients all seems well, but as clients are spawned, the spawning rate is falling.
For example when I have 20 clients running, the next one may take 3-5secs to be spawned.
To launch, I use: locust -f test_https.py -H https://www.google.com.
40 clients rate: 1cl/sec
Last run I did, I saw a delay of 7 sec between clients 46 and 47
I did add a profiler to locust, the differents method of sslSockets seems to take lots of time.
Is it a known problem ? Maybe i am just missing a patch or something (I hope so, but...)
About my system, I have the same problem in Ubuntu or Centos7, I am running on a pyenv environment, with python 2.7.10. I am working off the following tags:
it's not a problem, but rather the way secure communication works. There is overhead to encryption and cipher negotiation vs. sending plaintext requests. I would expect to see some level of degradation.
Hi all,
Since a few week, the server i am running against is upping it's security, they enabled HTTPS, I have been asked to run my tests over HTTPS, to load this new 'feature'.
On Http, I have been able to load a 1000Clients at a Client per sec rate without any problem.
Once I enable Https, for the few first clients all seems well, but as clients are spawned, the spawning rate is falling.
For example when I have 20 clients running, the next one may take 3-5secs to be spawned.
Just in case, to be sure the problem is not on our server,
I did a quick locust file trying to do gets on a few public servers, which can be ddl at the following link.
https://drive.google.com/file/d/0Bwj-3cp-XeYcY19CbnFuU2VKR3M/view?usp=sharing
To launch, I use: locust -f test_https.py -H https://www.google.com.
40 clients rate: 1cl/sec
Last run I did, I saw a delay of 7 sec between clients 46 and 47
I did add a profiler to locust, the differents method of sslSockets seems to take lots of time.
ncalls tottime percall cumtime percall filename:lineno(function)
1357 37.451 0.028 37.451 0.028 {method 'read' of '_ssl._SSLSocket' objects}
92 18.705 0.203 18.705 0.203 {method 'do_handshake' of '_ssl._SSLSocket' objects}
Is it a known problem ? Maybe i am just missing a patch or something (I hope so, but...)
About my system, I have the same problem in Ubuntu or Centos7, I am running on a pyenv environment, with python 2.7.10. I am working off the following tags:
If anyone has an idea as to how I can get out of this problem, i am all ear :)
Many thanks,
Cedric.
Just in case, the full link to the profiler, if s/o can understand it better than me:
https://drive.google.com/file/d/0Bwj-3cp-XeYcWWVPWnREeWdSMkU/view?usp=sharing
The text was updated successfully, but these errors were encountered: