Pycurl sometimes take up to 80%~100% CPU, causing congestion

StackOverflow https://stackoverflow.com/questions/21157813

  •  28-09-2022
  •  | 
  •  

Question

I'm using pycurl in a test application to stress my application via REST. I notice if I just send 1000 messages per second, pycurl would start taking up alot of CPU and sometimes return an error message.

pycurl.error: (7, 'Failed to connect to 127.0.0.1: Cannot assign requested address') Command exited with non-zero status 1

I looked online abit and noticed other people have noticed high CPU utilization as well. I'm wondering is there anything I can do, or configure? Maybe there's a way to modify how pycurl binds to the socket?

Thanks Morgan

Was it helpful?

Solution

You are probably leaking socket file descriptors. Make sure you are closing all connections and files you read/write from/to. You can get a reference of pycurl error codes at:

http://curl.haxx.se/libcurl/c/libcurl-errors.html

Also you might try to set CURLOPT_VERBOSE option, to get more debug information.

Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top