From 3a5e953ddcd5469667b4055729435294097c0ce3 Mon Sep 17 00:00:00 2001 From: Dorota Wojcik Date: Thu, 11 Sep 2025 08:10:33 +0200 Subject: [PATCH 1/3] pg maintenance downtime --- docs/platform/concepts/maintenance-window.md | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/docs/platform/concepts/maintenance-window.md b/docs/platform/concepts/maintenance-window.md index 9d6fa5626..a1d238b6e 100644 --- a/docs/platform/concepts/maintenance-window.md +++ b/docs/platform/concepts/maintenance-window.md @@ -1,5 +1,6 @@ --- -title: Service maintenance, updates and upgrades +title: Service maintenance, updates, and upgrades +sidebar_label: Service maintenance --- import ConsoleLabel from "@site/src/components/ConsoleIcons" @@ -35,10 +36,9 @@ See [Upgrade PostgreSQL to a major version][pg]. ### MySQL, PostgreSQL and Caching updates -For **MySQL®**, **PostgreSQL®** and **Caching**, the maintenance -process usually lasts around several seconds. The downtime comes from old -master stopping itself in a controlled manner and new master executing -promotion sequence after this. +For **MySQL®**, **PostgreSQL®** and **Caching**, the maintenance process should take no +more than 15 seconds. The downtime comes from old master stopping itself in a controlled +manner and new master executing promotion sequence after this. Once the promotion is complete the old master node starts forwarding requests to the new master node so the From af0229f87aafe1a078c05f4bd39cc7dc75b9e5c8 Mon Sep 17 00:00:00 2001 From: Dorota Wojcik Date: Thu, 11 Sep 2025 11:07:30 +0200 Subject: [PATCH 2/3] feedback --- docs/platform/concepts/maintenance-window.md | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/docs/platform/concepts/maintenance-window.md b/docs/platform/concepts/maintenance-window.md index a1d238b6e..3ff7c7381 100644 --- a/docs/platform/concepts/maintenance-window.md +++ b/docs/platform/concepts/maintenance-window.md @@ -37,14 +37,14 @@ See [Upgrade PostgreSQL to a major version][pg]. ### MySQL, PostgreSQL and Caching updates For **MySQL®**, **PostgreSQL®** and **Caching**, the maintenance process should take no -more than 15 seconds. The downtime comes from old master stopping itself in a controlled -manner and new master executing promotion sequence after this. +more than 15 seconds. The downtime comes from the old primary stopping itself in a +controlled manner and the new primary executing a promotion sequence after this. Once the promotion is complete the old -master node starts forwarding requests to the new master node so the +primary node starts forwarding requests to the new primary node so the service is accessible before DNS updates are propagated, though clients -that end up reconnecting to the old master node will see additional -disconnection once the old master is permanently retired. +that end up reconnecting to the old primary node will see additional +disconnection once the old primary is permanently retired. ### Apache Kafka and OpenSearch updates From 0c1c41e329f3857074972c876b28ac90a925e91e Mon Sep 17 00:00:00 2001 From: dorota <114921900+wojcik-dorota@users.noreply.github.com> Date: Tue, 28 Oct 2025 14:36:41 +0100 Subject: [PATCH 3/3] Update maintenance window documentation for Valkey Signed-off-by: dorota <114921900+wojcik-dorota@users.noreply.github.com> --- docs/platform/concepts/maintenance-window.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/platform/concepts/maintenance-window.md b/docs/platform/concepts/maintenance-window.md index 3ff7c7381..d0167cde0 100644 --- a/docs/platform/concepts/maintenance-window.md +++ b/docs/platform/concepts/maintenance-window.md @@ -34,9 +34,9 @@ Major service upgrades are triggered manually. A manually triggered upgrade star See [Upgrade PostgreSQL to a major version][pg]. ::: -### MySQL, PostgreSQL and Caching updates +### MySQL, PostgreSQL and Valkey updates -For **MySQL®**, **PostgreSQL®** and **Caching**, the maintenance process should take no +For **MySQL®**, **PostgreSQL®** and **Valkey**, the maintenance process should take no more than 15 seconds. The downtime comes from the old primary stopping itself in a controlled manner and the new primary executing a promotion sequence after this.