The recent issue was due to a toolkit update that included iptables changes. This resulted in certain IP blocking within the cluster and on 2 nodes outside the cluster. As a consequence, Patroni detected the cluster as unhealthy and attempted to identify a healthy candidate for switchover, but was unable to do so, leading to a Leader restart.
Subsequently, resiliency was partially impacted, and the 2 affected nodes were later brought back in sync under CHANGE_NUMBER123.
0 comments:
Post a Comment