As the title suggests, is it a specification that Rails 5+Puma+Nginx configuration releases memory at a fixed time (around 57 minutes per hour)?
The main versions used are as follows:
There are three servers and they are configured as follows:
I thought it was probably Puma's memory, so I checked if Puma has such a function, but I couldn't check the specifications.(When memory is released, Puma's worker process ID changed, so Puma's worker seems to have restarted.)
Below is a graph of Memory Used% in NewRelic, where the blue and red lines are running Puma, and you can see that memory usage drops sharply around 57 minutes per hour as described above.
By the way, I don't use puma_worker_killer.
So I asked you, is this the specification of Puma, Nginx (or Rails)?If this is the specification, please let me know if there is a document with the specification.ruby-on-rails nginx
Regarding this matter, I was rotating the logs of the Rails application in the Linux OS logrotate, but it was because I restarted the puma during the logrotate (because puma will not output the logs to the new output destination)
© 2022 OneMinuteCode. All rights reserved.