Description of problem: etcd has various runtime configurations. Currently, the only way to adjust these configurations is to manipulate the etcd-member.yaml for the static POD. But by doing so because this config is monitored by MCO will result in eventually restarting the node. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. adjust runtime configuration 2. 3. Actual results: etcd picks up runtime configuration changes and then MCO reboots nodes Expected results: etcd picks up runtime configuration Additional info:
hello Sam, I found that etcd-member.yml have been managed by machineconfig, do you have any suggestion how to verify this bug? thx
generate-etcd-conf.sh works well, but its pre-requirist is install 'urlencode'. I'm not sure if it's acceptable for customer, or we need to doc it, anyway, this way looks like have not problem. # urlencode bash: urlencode: command not found... Install package 'gridsite-clients' to provide command 'urlencode'? [N/y] y # ./generate-etcd-conf.sh etcd.conf.new etcd.conf.machineconfig MachineConfig is now available etcd.conf.machineconfig # oc get machineconfig NAME GENERATEDBYCONTROLLER IGNITIONVERSION CREATED 00-master 4.1.0-201905091432-dirty 2.2.0 25h .............. 99-etcd-conf 2.2.0 7s