Thread: Patroni set up

I am looking to DR set up patroni. For 5 node set up

Mumbai: Node 1: postgres+etcd+patroni 

Mumbai: Node 2: postgres+etcd+patroni

Chennai: Node 3: postgres+etcd+patroni 

Mumbai: haproxy+keepalived

Chennai: haproxy+keepalived

Will this work? And I want to know in which node should I take daily backup. Meaning- in which node I should configure pgbackrest backup

Hi, 

In my opinion this 5-node set up is very elaborate and should work , which is distributed different across different seismic zone (across Mumbai (primary DC) and Chennai (DR site))  . Probably below mentioned parameters were factored when designing the system

1. network latency between Mumbai and Chennai
2. application switch over in case of fail-over and the downtime.

Thanks 

Dinesh Nair

---

From: kaido vaikla <kaido.vaikla@gmail.com>
Sent: Wednesday, July 2, 2025 2:28 PM
To: Raj <rajeshkumar.dba09@gmail.com>
Cc: Pgsql-admin <pgsql-admin@lists.postgresql.org>
Subject: Re: Patroni set up

 

You don't often get email from kaido.vaikla@gmail.com. Learn why this is important

Caution: This email was sent from an external source. Please verify the sender’s identity before clicking links or opening attachments.

Hi, 
Based on my opinion and setups.
I would keep etcd cluster on separate nodes and separate data centre.
Mumbai down means whole pg patroni cluster is down, because you dont have an etcd quorum. And other reasons, why create separet etcd cluster.

Take a backup from leader. Leader donw means no replicas. Replica down does not mean leader down.

br
Kaido

On Wed, 2 Jul 2025 at 10:38, Raj <rajeshkumar.dba09@gmail.com> wrote:

I am looking to DR set up patroni. For 5 node set up

Mumbai: Node 1: postgres+etcd+patroni 

Mumbai: Node 2: postgres+etcd+patroni

Chennai: Node 3: postgres+etcd+patroni 

Mumbai: haproxy+keepalived

Chennai: haproxy+keepalived

Will this work? And I want to know in which node should I take daily backup. Meaning- in which node I should configure pgbackrest backup

Hi Raj, 

By default pgBackRest will take the backup from primary, but we can configure it take from stand by enabling the backup-standby=y.

Hope it answers your question!!

Thanks 

Dinesh Nair

---

From: Raj <rajeshkumar.dba09@gmail.com>
Sent: Wednesday, July 2, 2025 4:07 PM
To: DINESH NAIR <Dinesh_Nair@iitmpravartak.net>
Cc: kaido vaikla <kaido.vaikla@gmail.com>; Pgsql-admin <pgsql-admin@lists.postgresql.org>
Subject: Re: Patroni set up

 

You don't often get email from rajeshkumar.dba09@gmail.com. Learn why this is important

Caution: This email was sent from an external source. Please verify the sender’s identity before clicking links or opening attachments.

Hi Dinesh, now with 5 node which server to take backup

On Wed, 2 Jul 2025, 15:49 DINESH NAIR, <Dinesh_Nair@iitmpravartak.net> wrote:

Hi, 

In my opinion this 5-node set up is very elaborate and should work , which is distributed different across different seismic zone (across Mumbai (primary DC) and Chennai (DR site))  . Probably below mentioned parameters were factored when designing the system

1. network latency between Mumbai and Chennai
2. application switch over in case of fail-over and the downtime.

Thanks 

Dinesh Nair

---

From: kaido vaikla <kaido.vaikla@gmail.com>
Sent: Wednesday, July 2, 2025 2:28 PM
To: Raj <rajeshkumar.dba09@gmail.com>
Cc: Pgsql-admin <pgsql-admin@lists.postgresql.org>
Subject: Re: Patroni set up

 

You don't often get email from kaido.vaikla@gmail.com. Learn why this is important

Caution: This email was sent from an external source. Please verify the sender’s identity before clicking links or opening attachments.

Hi, 
Based on my opinion and setups.
I would keep etcd cluster on separate nodes and separate data centre.
Mumbai down means whole pg patroni cluster is down, because you dont have an etcd quorum. And other reasons, why create separet etcd cluster.

Take a backup from leader. Leader donw means no replicas. Replica down does not mean leader down.

br
Kaido

On Wed, 2 Jul 2025 at 10:38, Raj <rajeshkumar.dba09@gmail.com> wrote:

I am looking to DR set up patroni. For 5 node set up

Mumbai: Node 1: postgres+etcd+patroni 

Mumbai: Node 2: postgres+etcd+patroni

Chennai: Node 3: postgres+etcd+patroni 

Mumbai: haproxy+keepalived

Chennai: haproxy+keepalived

Will this work? And I want to know in which node should I take daily backup. Meaning- in which node I should configure pgbackrest backup

On Wed, 2 Jul 2025 at 13:19, DINESH NAIR <Dinesh_Nair@iitmpravartak.net> wrote:

Hi, 

In my opinion this 5-node set up is very elaborate and should work , which is distributed different across different seismic zone (across Mumbai (primary DC) and Chennai (DR site))  . Probably below mentioned parameters were factored when designing the system

1. network latency between Mumbai and Chennai
2. application switch over in case of fail-over and the downtime.

Thanks 

Dinesh Nair

If  Mumbai (primary DC) is down then is down pg+patroni on Chennai. etcd quurom is lost
br
Kaido