Kubernetes memory requests don't do what you think (until you enable MemoryQoS)
Kubernetes memory management relies on cgroups to allocate resources effectively. By default, memory requests are only utilized by the scheduler, which can lead to processes being reclaimed even if they stay within their declared limits. The MemoryQoS feature, still in alpha, allows for better memory guarantees but requires specific kernel versions and configurations to function properly.
- ▪Kubernetes uses cgroups for resource management, particularly focusing on cgroups v2 for memory management.
- ▪In the default configuration, memory requests are only used by the scheduler and not passed to the kernel as memory.min.
- ▪The MemoryQoS feature allows kubelet to map memory requests to memory.min, ensuring a minimum memory allocation for processes.
Opening excerpt (first ~120 words) tap to expand
(function(e,t,n,s,o){e[s]=e[s]||[],e[s].push({"gtm.start":(new Date).getTime(),event:"gtm.js"});var a=t.getElementsByTagName(n)[0],i=t.createElement(n),r=s!="dataLayer"?"&l="+s:"";i.async=!0,i.src="https://www.googletagmanager.com/gtm.js?id="+o+r,a.parentNode.insertBefore(i,a)})(window,document,"script","dataLayer","GTM-PM8VLKVT")<iframe src="https://www.googletagmanager.com/ns.html?id=GTM-PM8VLKVT" height=0 width=0 style=display:none;visibility:hidden></iframe>HomeBlogKubernetes Management: Memory Request and Limit in Practice April 19, 2026 | 7 min ReadKubernetes Management: Memory Request and Limit in PracticeThis is the second part of an article series about resource management in Kubernetes.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Roszigit.