Probably thinking I began to deal with the problem right now describing the problem might be a bit chaotic, but there is a possible approach to meet the people I want to help this pit
Description: springboot project, which docker mirror operation, see docker log, after the success of the project started, separated by a minute or so he would automatically restart, and then cause the interface to access the site when nginx reported 502 gateway Han, there are two servers, a file server, run the code is very simple upload and download files, and verify the token, the other running java applications, both servers are mirrored in the update project, run after run into this problem, very strange.
Then I ended up project jar packets directly java -jar xxx.jar, which runs directly in the application server, then the card can not continue in some places to start, or the display has been killed
We suspect that the project memory leak, causing the system to kill the process, then replaced before docker image, the result is the same effect, still have this problem, suspect nginx configuration changes, and changed still useless, then also became Ali cloud own server running the latest image, without any problems, completely normal, run jar wrapped in Ali cloud which is also normal, so the suspect could be a system where changes cause this problem, so we looked at the top, found
cpu usage is too high, do not know there is a very normal process, Baidu a bit, this cornd command is performed regularly wisdom, so find ways to solve it (ps: And this is not our equipment, the environment did not get online these)
First check the status: systemctl status crond
Hardy then configure which command: crontab -e
I 0 is changed, before he was *, change it to 0, then save and exit
Then stop the service: systemctl stop crond
Finally, look at the top, if there is this process exists, kill off directly, and then ojbk the pit dead, causing me to toss most of the day