4 ms·
I used many times and sometimes I used py-spy. That helped me to improve several projects where people told me there were network problems but it was actually n
by motbus3 2y ago
I used many times and sometimes I used py-spy. That helped me to improve several projects where people told me there were network problems but it was actually not
- alfons_foobar 2y agokudos for actually looking what the problem is. As a (former) NetEng, it bothers me to no end that so many people claim "it's the network" when their application is slow / broken, without understanding the actual problem.
- embeng4096 2y ago+1 for py-spy. I would love to try out Scalene but the application I have does something funky with pickling for multiprocessing, breaking Scalene in the process (my fault, not Scalene's, I'll be looking into that). Py-spy worked, including catching all the sub-processes. Feeding the py-spy JSON output into https://speedscope.app https://speedscope.app makes for a very easy way to profile in a time crunch if you don't have time to get familiar with CLI tools or can't install stuff locally but have a browser and internet connection.
- emeryberger 2y agoScalene author here - please file an issue!
- embeng4096 2y agoI will! I'll try to pare down my customer's code into a minimal example I can post.
- motbus3 2y agoScalene is great too. I actually use to find some memory allocation problems that were quite hard do pin.
- carlmr 2y ago>where people told me there were network problems but it was actually not Always ask if they assume it's network bound or they have measurements. Measurements may sometimes be wrong, but assumptions are more often wrong than right in performance engineering.