OmarElawady

Had the same issue for two workloads on the node "A7FmQZ72h7FzjkJMGXmzLDFyfyxzitDZYuernGG97nv7". Workloads: [482883](https://explorer.testnet.grid.tf/api/v1/reservations/workloads/482883) [482995](https://explorer.testnet.grid.tf/api/v1/reservations/workloads/482995)

GPUs can be attached to a VM using `cloud-hypervisor` by unbinding it from its driver and then bind it to `vfio` driver as described [here](https://github.com/cloud-hypervisor/cloud-hypervisor/blob/main/docs/vfio.md). I have a nvidia GPU,...

The limit is 256MB on zos and it only has the zdb process in the container (on which the limit is applied).

It's possible that it's a cache problem and the memory is allocated as a cache and not directly by zdb. From the OOM logs, the vm (virtual memory) used is...

hashmem also OOMs, zdb process restarts occasionally using 32M as datasize but not because of OOM (no dmesg logs, and container logs are truncated, segfault?). Not reproducible with me locally...

The zdb restarting occasionally is the latest with 32MB (not sure why as the logs are truncated). The memory limit is set on the node eventually using cgroups. But wasn't...