Re: [jupyter] Load Testing JupyterHub

2017-02-16 Thread Ryan Lovett
Hi all, Sorry for not noticing this thread earlier. Yes, Yuvi setup a selenium cluster to test a KubeSpawner-based deployment. He got as high as 1000 simultaneous active users before there was a problem, and then that was with the KubeSpawner itself. Ryan On Thu, Feb 16, 2017 at 2:00 AM, DVD

Re: [jupyter] Load Testing JupyterHub

2017-02-16 Thread Brian Granger
In general, the bottleneck is always going to be the N_students*resources_per_student. That varies widely by what the students are doing. In our data science courses, we give each students 2GB RAM and approximately 4 students per CPU core. The CPU stuff isn't typically the issue - RAM *always*

Re: [jupyter] Load Testing JupyterHub

2017-02-16 Thread DVD PS
On Friday, 2 September 2016 23:55:13 UTC+1, Fernando Perez wrote: > > On Fri, Sep 2, 2016 at 3:49 PM, Thorin Tabor > wrote: > >> Hello, >> >> We've been operating a JupyterHub instance for some time, and are now >> looking to gain some insight into the maximum

Re: [jupyter] Load Testing JupyterHub

2016-09-02 Thread Fernando Perez
On Fri, Sep 2, 2016 at 3:49 PM, Thorin Tabor wrote: > Hello, > > We've been operating a JupyterHub instance for some time, and are now > looking to gain some insight into the maximum number of users or notebooks > that our current hardware can handle. Basically we want