Closed GrumpyMeow closed 1 month ago
When trying to get more detail information for this issue i noticed that atop gave a error message about the memory usage. As it's a container, i was able to give it more memory. I never suspected this to be an issue as Proxmox didn't indicate that memory was an issue. It's a known issue that Proxmox sometimes is unsable to give correct stats. The Proxmox screenshot indicated it never used more that 4gb. But in actuallity it was using 100% of memory.
Found:
The generated VM image is temporarily written to the /tmp folder in my NixOS LXC. To reduce disk-wear i had this configured with boot.tmp.useTmpfs = true;
. So the VM image was being written to memory.
Hi! When i generate an image the machine becomes very unresponsive and other processes on the machine starve. Eventually when the image is generated, the machine becomes responsive again and the processes do start to work again. I report this as an issue as it's not common nowadays to be able to have a single process to starve the other running processes.
The diskio indicates that the machine is busy doing a diskread at it's maximum speed steadily (1.45G per second) for a long time.
Is it maybe possible to implement some throttling or set the process more "nice"? Having a more responsive machine would also allow me to check what it's currently doing, as the generate i'm currently running is going on for a very long time at full-speed diskread for halve-hour. It's not that big of a drive.
Details... The machine i'm running NixOS on is a ProxmoxLXC container. The container has been configured to use 12 of the 16 cores. 8GB of memory. (no swap) Command i was running:
nixos-generate -c mynewvm.nix -f proxmox