fanlix

Results 16 comments of fanlix

by #4560 , `overcommitGuestOverhead: true` help this case?

So, additional 300M(calculated by [the code](https://github.com/kubevirt/kubevirt/blob/18b90f3ff8e174a7a2365f296154ac00329bfe9a/pkg/virt-controller/services/renderresources.go#L272)) is added by `overcommitGuestOverhead: false` config. This is enough for normal kubevirt usage. But VM of heavy load cost more memory than expected. Causing...

top status of an oom ## 1, vm running stable. * the **hv.os** top shows ![hv-mem1](https://user-images.githubusercontent.com/987637/176846695-83f08103-03b8-4698-9c27-16c4104e3fd5.png) * the **vm.os** top shows ![vm-mem1](https://user-images.githubusercontent.com/987637/176846662-c8be35e3-55f7-49ee-b212-b96462ced89b.png) ## 2, kill & restart app2. * **hv.os**...

yes, this oom happend in HOST side. then how to avoid? * I already decrease ram Overcommit from 150% to 110%. should I decreate to 90% ? * should I...

About the memory-hole problem, any config to force kubevirt pre-alloc whole 64G memory? then HOST donnot need any dynamic complex memory logic, just handle one piece of pure physical ram....

Recently my heavy loading VMs oom-crash everyday. For example a VM running 20G app, I already alloc 72G ram, but skill oom-killed. I really need some method to fix or...

@guangbochen thanks very much. just trying editing k8 yml for cgroup limit.

By increate resoved memory from 100M(default) to 256M, cgroup-oom-kill not seen for 3 days now. In loadtest, both guest-os-oom & guest-os-swap triggered. _But i dont know how much resoved-memory is...

the memory calculate code is here: https://github.com/harvester/harvester/blob/master/pkg/webhook/resources/virtualmachine/mutator.go#L171-L182 different resovle-memory setups result in vmi config: (overcommit=120%) * default 100M ![resv-100M](https://user-images.githubusercontent.com/987637/183404715-fc508abf-ecf9-4534-9e8d-9053d873587d.PNG) * force 0 ![resv-0](https://user-images.githubusercontent.com/987637/183404218-c8fb1c14-9806-4af4-99bb-5f7f08ca01ed.PNG) * setup 256M ![resv-256M](https://user-images.githubusercontent.com/987637/183404287-58c9c765-b38b-4fa5-b4f1-16b7d76d1f69.PNG)

cgroup-oom-kill not happened for 3 weeks. My setup for resolved memory now: * tiny vm = default 100M * normal vm = 256M * ram>64G vm = 512M (maybe overkill,...