I'm not sure why Restart=on-failure is the default. On my systems, I always default to Restart=always. I don't see value in shying away from a hammer approach when you need something to always be running.
$0.02, On Tue, Jan 16, 2018 at 2:01 PM, hskarlupka < heath.skarlu...@icecube.wisc.edu> wrote: > All, > > I'm running a 4 node CentOS 7 Atomic Kubernetes Cluster. I ran into an > issue over the weekend where two of my nodes stopped running flanneld. > "Systemctl status flanneld" showed that the processes had an exit status of > 0 and were not running. The Unit file sets "Restart=on-failure" instead of > "Restart=always". Is there a reason for not setting it to > "Restart=Always"? The flanneld process seems to exit 0 more often on idle > nodes. It looks like CoreOS's example file uses "Restart=Always": > https://github.com/coreos/coreos-overlay/blob/master/app- > admin/flannel-wrapper/files/flanneld.service > > Thank you for the info! > > Heath > > -- Ben Breard Sr Technology Product Manager - Linux Containers Mobile: 972-816-9081
