ETCD no leader when nodes have problem connect with leader node #16502
Replies: 3 comments 1 reply
-
Hey @EdithChenLi - Thanks for raising this question. This won't be what you want to hear however etcd |
Beta Was this translation helpful? Give feedback.
-
thanks for the response @jmhbnz . I know this version is bit old, but we installed the ETCD using Podman image which latest version is etcd:3.2.32 |
Beta Was this translation helpful? Give feedback.
-
@jmhbnz For your question, I did not use > 3.4 version before, not quite sure about if same issue happens. As ETCD doc, seems there is lease and leaner nodes setup which should help fix this issue |
Beta Was this translation helpful? Give feedback.
-
ETCD 3-nodes connection based on certificates, when I stop leader node service or server, other 2 nodes will immediately promote 1 of them to be leader. But I found if others 2nodes have connect problem with leader node, like certificate key lost/expired or network delay, no new leader will promote. The PostgreSQL cluster will become read-only after 1-3 mins.
ETCD version is etcd:3.2.32(podman image). is this expected?
Failed to get the status of endpoint xxxx:2379(rpc error: code = internal desc=connection errro: desc = "transport: authentication handshake failed: remote error: tls: internal error")
ENDPOINT ID VERSION DB SIZE IS LEADER RAFT TERM RAFT INDEX
xxxxxx:2379 xxxxxxx 3.2.32 156kb false 18 142691
xxxxxx:2379 xxxxxxx 3.2.32 160kb false 18 142691
Beta Was this translation helpful? Give feedback.
All reactions