On Mon, Aug 25, 2025 at 10:16 AM Dave Voutila <d...@sisu.io> wrote: > > Mike Larkin <mlar...@nested.page> writes: > > > On Sun, Aug 24, 2025 at 02:05:26PM -0400, Kevin wrote: > >> Hey gang, > >> > >> Have some VMs that take longer than the 30 seconds to shutdown that's > >> defined in vmd.h > >> > >> #define VM_SHUTDOWN_TIMEOUT 30 > >> > >> The VMs in question all have multiple disks and tons of files that > >> httpd has a proper kung-fu grip on, so... > >> > > OP: How long does httpd take to stop in your case?
When it's busier than a one-legged man in an ass kicking contest, about a minute. > > >> when you issue > >> > >> vmctl stop 23 > >> > >> VMD gives up waiting for the VM to shut down and just kills it anyway... > >> > >> which just leads to tears and a painfully long ass fsck. > >> > >> I seem to remember there's a knob -- somewhere -- that lets you define > >> how long VMD should wait, but I sure can't find the damned thing. > >> > >> Anyone know where this knob lives? > >> > >> Thanks, > >> Kevin > >> > > > > if we were to change it (not saying we will), how much time do you propose? > > My backlog of neverending vmd improvement ideas involves a re-write of > vmmci(4) and the device side. I have ideas for what could improve this > and at least let the guest-side maybe better communicate progress so the > big hammer approach could be softened a bit. I can imagine. and, it's already worlds better than other options like qemu.
