**High Priority** repmgr with multi slaves replication solution is not working in k8s environment

[Bhagi33]

Hi Team,

On kubernetes i am deploying postgres with
1 master(postgres-master-0),
1 witness(witness-slave-0-0)
2 slaves: slave1(witness-slave-1-0) & slave2(witness-slave-2-0)

But repmgr replication solution is not working. when i register 2 standby's , they are trying to run as primary .
i am attaching the screenshots and its configuration parameters. please guide me to fix the issue.

-> Below configuration is of Master and standby's

cat /persistent/postgres/postgresql/pg_hba.conf

local replication repmgr trust
host replication repmgr 127.0.0.1/32 trust
host replication repmgr 16.0.0.0/16 trust

local repmgr repmgr trust
host repmgr repmgr 127.0.0.1/32 trust
host repmgr repmgr 16.0.0.0/16 trust
local postgres embahar trust
host replication repmgr 0.0.0.0/0 trust
host repmgr repmgr 0.0.0.0/0 trust
host postgres grafmon 0.0.0.0/0 md5

witness-slave-1-0:/home/postgres$ cat /persistent/postgres/postgresql/postgresql.conf
listen_addresses = '*'
shared_preload_libraries = 'repmgr,pg_cron'
max_wal_senders = 10
max_replication_slots = 10
wal_level = 'replica'
hot_standby = on
archive_mode = on
archive_command = '/bin/true'

--> Witenss configuration

cat /persistent/postgres/postgresql/pg_hba.conf
local replication repmgr trust
host replication repmgr 127.0.0.1/32 trust
host replication repmgr 16.0.0.0/16 trust

local repmgr repmgr trust
host repmgr repmgr 127.0.0.1/32 trust
host repmgr repmgr 16.0.0.0/16 trust
local postgres embahar trust
host replication repmgr 0.0.0.0/0 trust
host repmgr repmgr 0.0.0.0/0 trust
host postgres grafmon 0.0.0.0/0 md5

witness-slave-0-0:/home/postgres$ cat /persistent/postgres/postgresql/postgresql.conf
listen_addresses = '*'
shared_preload_libraries = 'repmgr,pg_cron'
witness-slave-0-0:/home/postgres$

[gonzalemario]

Hi, have you made any progress? It seems there are connection issues between the cluster.

Although repmgr should work, there are way more important stuff to take care before worrying about primary/standby on k8s or containers. I'd personally jump into CNPG to handle postgres life-cycle. We don't have plans to support containers in repmgr so you would be working towards something that's untested. Moreover, if you use some k8s operator like CNPG, you don't need repmgr or patroni.

I'll keep this open for a few days in case you want more feedback but IMO, we might close this as "not planned", sorry.