aledbf Wrote: ------------------------------------------------------- > We used jemalloc in the past, but that approach also introduces > different issues like not being able to use third-party monitory > agents like dynatrace. Was it for the same reason as me you tried jemalloc? did you find any other solutions? Also, did you set it up via LD_PRELOAD, or how did you set it up?by fredr - Nginx Mailing List - English
I guess you are right. The main reason I want to scale on memory rather than number of connections, is that we wouldn't have to calculate how many connections a node can handle. Eg, if we change the memory size of each node, we also have to update the automatic scaling metric, or lets say there is a new version of nginx that uses less memory per connection, then we would have to re-calibrate the sby fredr - Nginx Mailing List - English
Maxim Dounin Wrote: ------------------------------------------------------- > > Whether or not allocated (and then freed) memory will be returned > to the OS depends mostly on your system allocator and its > settings. That is very interesting! I had no idea, thanks! Maxim Dounin Wrote: ------------------------------------------------------- > On Linux with standardby fredr - Nginx Mailing List - English
Hi all, We are using the kubernetes nginx-ingress for websocket connections in front of one of our applications. We have added automatic scaling based on the resident memory, as that seems to be a good scaling metric when dealing with persistent connections. But we noticed that the memory seems to never be released, and thus it only scales up and never down. I've done some testing locally usby fredr - Nginx Mailing List - English