Recently we started getting 500 response status on our api end point and when we checked logs, we have errors coming from pymemcache code.
I have opened SO question as well https://stackoverflow.com/questions/79250060/pymemcache-oserror-errno-99-cannot-assign-requested-address but I think this is something either buggy in pymemcache or something that is not understood by me and thus used incorrectly.
We are using version pymemcache 4.0.0 and memcached for caching purpose. Our application is django based, so we are using the provided api cache.get() to access cache https://docs.djangoproject.com/en/5.1/topics/cache/#memcached
Does the pooled client not reuse existing connections by default? We see a huge number of ports in the wait_time state as it looks like pymemcache opens new connection to memcache server for every request we receive on our application and the ports are thus occupied.
cat /proc/net/tcp|grep 2BCB|grep " 03:"|wc -l
the output of above command keeps on increasing if we triggered huge number of requests on our application server.
Recently we started getting 500 response status on our api end point and when we checked logs, we have errors coming from pymemcache code.
I have opened SO question as well https://stackoverflow.com/questions/79250060/pymemcache-oserror-errno-99-cannot-assign-requested-address but I think this is something either buggy in pymemcache or something that is not understood by me and thus used incorrectly.
We are using version pymemcache 4.0.0 and memcached for caching purpose. Our application is django based, so we are using the provided api cache.get() to access cache https://docs.djangoproject.com/en/5.1/topics/cache/#memcached
Does the pooled client not reuse existing connections by default? We see a huge number of ports in the wait_time state as it looks like pymemcache opens new connection to memcache server for every request we receive on our application and the ports are thus occupied.
cat /proc/net/tcp|grep 2BCB|grep " 03:"|wc -lthe output of above command keeps on increasing if we triggered huge number of requests on our application server.