NodeMemoryMajorPagesFaults after upgrading to 9.0.1

[maazamalik]

Elasticsearch Version

9.0.1

Installed Plugins

No response

Java Version

bundled

OS Version

Linux <pod_name> 5.10.234-225.921.amzn2.x86_64 #1 SMP Tue Mar 25 21:34:29 UTC 2025 x86_64 x86_64 x86_64 GNU/Linux

Problem Description

Hi all,

We are running elasticsearch on kubernetes using eck operator.
Last week we upgraded to 9.0.1, since then we are getting major memory page faults on the node running the elasticsearch data node causing this alert to fire in prometheus
"NodeMemoryMajorPagesFaults Memory major page faults are occurring at very high rate."

Our indexing rate hasn't changed during or after the upgrade nor the deletion rate so we can't blame that. Can anyone explain why this could be happening?

As you can see in the screenshot, it only started happening from June 11th which was the date of upgrade

Steps to Reproduce

upgrade to 9.0.1

Logs (if relevant)

No response

EDIT:
some important points regarding our cluster setup

• Swap is disabled on the node
• bootstrap.memory_lock: true
• 3 Master nodes, 24 Data nodes
• Per Master node resources: (cpu req/limit: 2/none, memory req/limit: 13.5Gi/13.5Gi ),
• Per Data node resources (cpu req/limit: 4/none, memory req/limit: 29Gi/29Gi),
• Total data size in cluster: ~13TB

We also have 2 more clusters fairly similar in size and data and they also have the same issue

[applike-ss]

We are affected as well, see: https://discuss.elastic.co/t/increased-read-iops-usage-after-upgrade-from-8-18-0-to-9-0-1/377986/1

[elasticsearchmachine]

Pinging @elastic/es-distributed-indexing (Team:Distributed Indexing)

[maazamalik]

Any updates on this ?