Consul-server always restarts election and no cluster leader

hTangle · October 12, 2021, 9:29am

Hi,
I have four kubernetes nodes

kubectl get nodes
NAME            STATUS     ROLES                      AGE   VERSION
192.168.1.116   Ready      controlplane,etcd,worker   12h   v1.18.9-6+3d5bea383c722a
192.168.1.172   Ready   worker                     12h   v1.18.9-6+3d5bea383c722a
192.168.1.20    Ready      controlplane,etcd,worker   12h   v1.18.9-6+3d5bea383c722a
192.168.1.88    Ready      controlplane,etcd,worker   12h   v1.18.9-6+3d5bea383c722a
kubectl get pods -n consul -o wide|grep server
consul-server-0                  1/1     Running        0          45m   10.42.3.45    192.168.1.172   <none>           <none>
consul-server-1                  1/1     Running            0          45m   10.42.0.16    192.168.1.20    <none>           <none>
consul-server-2                  1/1     Running            0          46m   10.42.2.125   192.168.1.116   <none>           <none>

I do the following steps to reproduce this issue:

sudo systemctl stop docker in 192.168.1.116 : after doing that consul cluster work normally
sudo systemctl start docker in 192.168.1.116 and at the same time sudo systemctl stop docker in 192.168.1.172:

kubectl get pods -n consul -o wide|grep server
consul-server-0                  1/1     Terminating        0          45m   10.42.3.45    192.168.1.172   <none>           <none>
consul-server-1                  0/1     Running            0          45m   10.42.0.16    192.168.1.20    <none>           <none>
consul-server-2                  0/1     Running            0          46m   10.42.2.125   192.168.1.116   <none>           <none>

The consul cluster do not work.
For getting more log for what’s happening here, I just add some logs in GitHub - hashicorp/raft: Golang implementation of the Raft consensus protocol and rebuild consul-1.9.1
In consul-server-2:
Consul members as below

/ # consul members
Node             Address           Status  Type    Build     Protocol  DC   Segment
consul-server-1  10.42.0.16:8301   alive   server  1.9.1dev  2         pri  <all>
consul-server-2  10.42.2.125:8301  alive   server  1.9.1dev  2         pri  <all>

And consul-server-2 always restarts election

2021-09-29T02:44:30.385Z [WARN]  agent.server.raft: Election timeout reached, restarting election
    2021-09-29T02:44:30.385Z [INFO]  agent.server.raft: entering candidate state: node="Node at 10.42.2.125:8300 [Candidate]" term=92
    2021-09-29T02:44:30.385Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980650 address->=10.42.3.45:8300
    2021-09-29T02:44:30.385Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980651 address->=10.42.0.16:8300
    2021-09-29T02:44:30.385Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980652 address->=10.42.2.88:8300
    2021-09-29T02:44:30.385Z [INFO]  agent.server.raft: ask peer to vote:: id->=697db73f-6bb9-f717-1c24-81139c980650 address->=10.42.3.45:8300
Could not find address for server id 697db73f-6bb9-f717-1c24-81139c980650
ID: 697db73f-6bb9-f717-1c24-81139c980652 Name: consul-server-2 Addr: 10.42.2.125:8300
ID: 697db73f-6bb9-f717-1c24-81139c980651 Name: consul-server-1 Addr: 10.42.0.16:8300
    2021-09-29T02:44:30.386Z [WARN]  agent.server.raft: unable to get address for server, using fallback address: id=697db73f-6bb9-f717-1c24-81139c980650 fallback=10.42.3.45:8300 error="Could not find address for server id 697db73f-6bb9-f717-1c24-81139c980650"
    2021-09-29T02:44:30.386Z [INFO]  agent.server.raft: ask peer to vote:: id->=697db73f-6bb9-f717-1c24-81139c980651 address->=10.42.0.16:8300
    2021-09-29T02:44:30.973Z [ERROR] agent.server.raft: failed to make requestVote RPC: target="{Voter 697db73f-6bb9-f717-1c24-81139c980650 10.42.3.45:8300}" error="dial tcp <nil>->10.42.3.45:8300: i/o timeout"
    2021-09-29T02:44:31.986Z [WARN]  agent: Syncing node info failed.: error="No cluster leader"
    2021-09-29T02:44:31.986Z [ERROR] agent: failed to sync changes: error="No cluster leader"
    2021-09-29T02:44:36.743Z [WARN]  agent.server.raft: Election timeout reached, restarting election
    2021-09-29T02:44:36.743Z [INFO]  agent.server.raft: entering candidate state: node="Node at 10.42.2.125:8300 [Candidate]" term=93

and in consul-server-1

/ # consul members
Node             Address           Status  Type    Build     Protocol  DC   Segment
consul-server-0  10.42.3.45:8301   failed  server  1.9.1dev  2         pri  <all>
consul-server-1  10.42.0.16:8301   alive   server  1.9.1dev  2         pri  <all>
consul-server-2  10.42.2.125:8301  alive   server  1.9.1dev  2         pri  <all>

and logs

2021-09-29T02:44:22.433Z [INFO]  agent.server.raft: entering candidate state: node="Node at 10.42.0.16:8300 [Candidate]" term=92
    2021-09-29T02:44:22.433Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980650 address->=10.42.3.45:8300
    2021-09-29T02:44:22.433Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980651 address->=10.42.0.16:8300
    2021-09-29T02:44:22.433Z [INFO]  agent.server.raft: ask peer to vote:: id->=697db73f-6bb9-f717-1c24-81139c980650 address->=10.42.3.45:8300
Could not find address for server id 697db73f-6bb9-f717-1c24-81139c980650
ID: 697db73f-6bb9-f717-1c24-81139c980651 Name: consul-server-1 Addr: 10.42.0.16:8300
ID: 697db73f-6bb9-f717-1c24-81139c980652 Name: consul-server-2 Addr: 10.42.2.125:8300
    2021-09-29T02:44:22.433Z [WARN]  agent.server.raft: unable to get address for server, using fallback address: id=697db73f-6bb9-f717-1c24-81139c980650 fallback=10.42.3.45:8300 error="Could not find address for server id 697db73f-6bb9-f717-1c24-81139c980650"
    2021-09-29T02:44:22.920Z [INFO]  agent.server.serf.lan: serf: attempting reconnect to consul-server-0 10.42.3.45:8301
    2021-09-29T02:44:24.967Z [ERROR] agent: failed to sync changes: error="No cluster leader"
    2021-09-29T02:44:25.921Z [ERROR] agent.server.raft: failed to make requestVote RPC: target="{Voter 697db73f-6bb9-f717-1c24-81139c980650 10.42.3.45:8300}" error="dial tcp <nil>->10.42.3.45:8300: i/o timeout"
    2021-09-29T02:44:30.386Z [INFO]  agent.server.raft: requestVote: from : 10.42.2.125:8300="req term" %!s(uint64=92)="current term" EXTRA_VALUE_AT_END=92
    2021-09-29T02:44:30.386Z [INFO]  agent.server.raft: duplicate requestVote for same term: term=92
    2021-09-29T02:44:32.079Z [ERROR] agent: failed to sync changes: error="No cluster leader"
    2021-09-29T02:44:32.220Z [WARN]  agent.server.raft: Election timeout reached, restarting election
    2021-09-29T02:44:32.220Z [INFO]  agent.server.raft: entering candidate state: node="Node at 10.42.0.16:8300 [Candidate]" term=93

and a WARN message agent.server.raft: rejecting vote request since our last index is greater: candidate=10.42.2.125:8300 last-index=8174 last-candidate-index=8033

    2021-09-29T02:55:48.722Z [WARN]  agent.server.raft: Election timeout reached, restarting election
    2021-09-29T02:55:48.722Z [INFO]  agent.server.raft: entering candidate state: node="Node at 10.42.0.16:8300 [Candidate]" term=189
    2021-09-29T02:55:48.722Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980650 address->=10.42.3.45:8300
    2021-09-29T02:55:48.722Z [INFO]  agent.server.raft: current server:: id->=697db73f-6bb9-f717-1c24-81139c980651 address->=10.42.0.16:8300
    2021-09-29T02:55:48.722Z [INFO]  agent.server.raft: ask peer to vote:: id->=697db73f-6bb9-f717-1c24-81139c980650 address->=10.42.3.45:8300
Could not find address for server id 697db73f-6bb9-f717-1c24-81139c980650
ID: 697db73f-6bb9-f717-1c24-81139c980651 Name: consul-server-1 Addr: 10.42.0.16:8300
ID: 697db73f-6bb9-f717-1c24-81139c980652 Name: consul-server-2 Addr: 10.42.2.125:8300
    2021-09-29T02:55:48.723Z [WARN]  agent.server.raft: unable to get address for server, using fallback address: id=697db73f-6bb9-f717-1c24-81139c980650 fallback=10.42.3.45:8300 error="Could not find address for server id 697db73f-6bb9-f717-1c24-81139c980650"
    2021-09-29T02:55:48.800Z [INFO]  agent.server.raft: requestVote: from : 10.42.2.125:8300="req term" %!s(uint64=188)="current term" EXTRA_VALUE_AT_END=189
    2021-09-29T02:55:48.800Z [INFO]  agent.server.raft: req term: %!s(uint64=188)="is less than" EXTRA_VALUE_AT_END=189
    2021-09-29T02:55:52.546Z [ERROR] agent: failed to sync changes: error="No cluster leader"
    2021-09-29T02:55:53.212Z [ERROR] agent.server.raft: failed to make requestVote RPC: target="{Voter 697db73f-6bb9-f717-1c24-81139c980650 10.42.3.45:8300}" error="dial tcp <nil>->10.42.3.45:8300: i/o timeout"
    2021-09-29T02:55:56.134Z [INFO]  agent.server.raft: requestVote: from : 10.42.2.125:8300="req term" %!s(uint64=190)="current term" EXTRA_VALUE_AT_END=189
    2021-09-29T02:55:56.135Z [WARN]  agent.server.raft: rejecting vote request since our last index is greater: candidate=10.42.2.125:8300 last-index=8174 last-candidate-index=8033
    2021-09-29T02:55:56.135Z [INFO]  agent.server.raft: entering follower state: follower="Node at 10.42.0.16:8300 [Follower]" leader=
    2021-09-29T02:55:58.723Z [ERROR] agent.server.raft: failed to make requestVote RPC: target="{Voter 697db73f-6bb9-f717-1c24-81139c980650 10.42.3.45:8300}" error="dial tcp <nil>->10.42.3.45:8300: i/o timeout"
    2021-09-29T02:55:59.420Z [ERROR] agent: Coordinate update error: error="No cluster leader"
    2021-09-29T02:55:59.701Z [ERROR] agent: failed to sync changes: error="No cluster leader"
    2021-09-29T02:56:01.818Z [INFO]  agent.server.raft: heartbeat timeout: lastContact: %s HeartbeatTimeout: %s: 2021-09-29 02:39:09.13070157 +0000 UTC m=+1186.321000621=5s
    2021-09-29T02:56:01.818Z [WARN]  agent.server.raft: heartbeat timeout reached, starting election: last-leader=
    2021-09-29T02:56:01.818Z [INFO]  agent.server.raft: entering candidate state: node="Node at 10.42.0.16:8300 [Candidate]" term=191

From the log, consul-server-1’s last index is greater than consul-server-2, so consul-server-1 should be leader not consul-server-2.
But raft log shows that only two server in consul-server-1's configurations.latest.Servers which not contains consul-server-2, so the vote requests will not send to consul-server-2 and the consul-server-1 will never become a leader because of consul-server-0’s dead .

And consul-server-2 has all servers in configurations.latest.Servers and can start vote normally, but consul-server-2 cannot be leader.
Why consul-server-2 in consul-server-1's members but not in configurations.latest.Servers.

This issue can be reproduced by following the step mentioned before.

Topic		Replies	Views
Consul not selecting leader Consul	1	10	December 5, 2024
Failed leadership election with three node cluster in GKE (Consul v1.5.2) Consul	4	363	February 20, 2023
[K8s] consul-controller pod leader election lost Consul k8s , connect , service-mesh	1	1117	May 11, 2022
Unable to make fault-tolerant 5 node Consul server setup Consul k8s , raft , consul	5	413	November 14, 2022
Kubernetes Consul Cluster Outage Recovery Consul	0	240	August 31, 2022

Consul-server always restarts election and no cluster leader

Related topics