solved

[vitabaks]

Hi @emanfeah

The log you provided shows that the Patroni service has encountered problems connecting to your etcd nodes at addresses 10.0.30.53 , 10.0.30.8, and 10.0.30.223 (port 2379).

To solve this problem, I recommend starting by checking the availability and health of etcd nodes. Then make sure there are no connection problems between the nodes.

from db node:

ping 10.0.30.53
ping 10.0.30.8
ping 10.0.30.223

telnet 10.0.30.53 2379
telnet 10.0.30.8 2379
telnet 10.0.30.223 2379

on etcd node

etcdctl version
etcdctl member list
etcdctl endpoint health --cluster

sudo journalctl -u etcd -n 1000

[vitabaks]

it looks like the etcd cluster is healthy.

Are you using the latest version of the playbook from the master branch?

etcdctl version: 3.5.7 API version: 3.5

Since we switched to using the API v3 and in your logs I see that Patroni is trying to use the etcd API v2

Failed to get list of machines from http://10.0.30.223:2379/v2

Please show the configuration file /etc/patroni/patroni.yml

the etcd3 section should be declared there, not etcd

https://github.com/vitabaks/postgresql_cluster/blob/master/roles/patroni/templates/patroni.yml.j2#L33

[vitabaks]

PostgreSQL is not running because Patroni is not running due to etcd connection problems.

i use the master branch -- etcd3

An interesting situation, I don't have any ideas yet on how to reproduce it.

Please attach the archive of your playbook, I'll take a look at it.

[vitabaks]

@emanfeah I have not been able to reproduce the problem using your variable files (from mail).

TASK [patroni : Prepare PostgreSQL | make sure the postgresql log directory "/data1/var/log/postgresql" exists] ***
changed: [10.172.0.21]
changed: [10.172.0.20]
changed: [10.172.0.22]

TASK [patroni : Prepare PostgreSQL | make sure PostgreSQL data directory "/data1/var/lib/postgresql/14/patroni" exists] ***
changed: [10.172.0.22]
changed: [10.172.0.21]
changed: [10.172.0.20]

TASK [patroni : Prepare PostgreSQL | check that data directory "/data1/var/lib/postgresql/14/patroni" is not initialized] ***
ok: [10.172.0.21]
ok: [10.172.0.20]
ok: [10.172.0.22]

TASK [patroni : Prepare PostgreSQL | make sure the postgresql config files exists] ***
ok: [10.172.0.20]
ok: [10.172.0.21]
ok: [10.172.0.22]

TASK [patroni : Prepare PostgreSQL | generate default postgresql config files] ***
changed: [10.172.0.20]
changed: [10.172.0.21]
changed: [10.172.0.22]

TASK [patroni : Prepare PostgreSQL | make sure the data directory "/data1/var/lib/postgresql/14/patroni" is empty] ***
changed: [10.172.0.20] => (item=absent)
changed: [10.172.0.22] => (item=absent)
changed: [10.172.0.21] => (item=absent)
changed: [10.172.0.21] => (item=directory)
changed: [10.172.0.20] => (item=directory)
changed: [10.172.0.22] => (item=directory)

TASK [patroni : Start patroni service on the Master server] ********************
changed: [10.172.0.20]

TASK [patroni : Wait for port 8008 to become open on the host] *****************
ok: [10.172.0.20]

TASK [patroni : Check PostgreSQL is started and accepting connections on Master] ***
ok: [10.172.0.20]

TASK [patroni : Wait for the cluster to initialize (master is the leader with the lock)] ***
ok: [10.172.0.20]

[vitabaks]

I don't think so, since my tests passed successfully. There is another problem in your case in the interaction of Patroni with the etcd cluster, which I may not be aware of.

Apparently your environment is somewhat different from what I used for testing. Try to deploy the cluster without changing any variables other than addresses in inventory.

[emanfeah]

do you think the problem its in

 pg_hba:  # Add following lines to pg_hba.conf after running 'initdb'
    - host replication replicator 127.0.0.1/32 md5
    - host all all 0.0.0.0/0 md5

???? i think it must be a master host ..

[vitabaks]

These are the default values for initdb and we redefine them further.

You can check the contents of the pg_hba.conf file after deployment

https://github.com/vitabaks/postgresql_cluster/blob/master/roles/patroni/tasks/main.yml#L858 https://github.com/vitabaks/postgresql_cluster/blob/master/roles/patroni/templates/pg_hba.conf.j2

[vitabaks]

Try to deploy the cluster without changing any variables other than addresses in inventory.

Let me know if you managed to deploy the cluster with the default values.

[vitabaks]

well, now the problem is at an earlier stage - the etcd cluster does not start

please attach logs from all servers sudo journalctl -u etcd

[emanfeah]

its work thank you

[vitabaks]

its work thank you

Ok.

Can you provide more details about your deployment? I would really like to find a way to reproduce your problem.