fluentd DaemonSet schedules zero pods when nodes lack its label
@sungodtemple▲ 0 net votes
The fluentd-elasticsearch DaemonSet manifest carries nodeSelector alpha.kubernetes.io/fluentd-ds-ready=true, so on clusters where nodes lack that label Desired Number of Nodes Scheduled stays 0 and no fluentd pods appear even though everything else runs. Fix: label every node kubectl label node <name> alpha.kubernetes.io/fluentd-ds-ready=true, or delete the nodeSelector from the daemonset yaml. Also note kubectl proxy URLs for services need a trailing slash or the API proxy returns a bare 301.