One of our servers (no details can be given here in public) experienced some swapping and started to get into trouble. While looking at it, I did some curl calls to the pm port's /procs/ endpoint. These blocked for several minutes. I am not sure that anything is wrong with the code; this could have just been due to the swapping negatively affecting the process as a whole. But it might be a good idea to make some benchmark tests, which work like this:
- add some large N of processes to the pm, such as 10000
- benchmark how long it takes to add a new process with Start()
- benchmark how long it takes to remove with Done() and update with Status()
One of our servers (no details can be given here in public) experienced some swapping and started to get into trouble. While looking at it, I did some curl calls to the pm port's
/procs/endpoint. These blocked for several minutes. I am not sure that anything is wrong with the code; this could have just been due to the swapping negatively affecting the process as a whole. But it might be a good idea to make some benchmark tests, which work like this: