On Tue, Jan 05, 2021 at 12:49:29PM -0700, Tracey Emery wrote: > Hello tech@, > > Some of us have been having shutdown issues with our VMs on OpenBSDAms. > I tracked down the problem to too short of a timeout for the shutdown > event. > > If there are an additional 1 or 2 package daemons running on the instance, > the timeout triggers before the VM has shutdown the package daemons and > properly synced the disks, resulting in a dirty startup. > > I've increased the timeout to 2 minutes instead of 30 seconds. My test > VM on my laptop with 7 additional package daemons succeeded in 60 > seconds, but that might not be fast enough for slower disks. > > Am I being conservative enough with this number? Should it be another > minute or two? > > Thoughts? Ok? > > -- > > Tracey Emery > > diff 7a6bb14936050379800deb10d4a137c4d2d4a3c4 /usr/src > blob - 9a64973ab998accb810d56c386c1bb92c204ab20 > file + usr.sbin/vmd/virtio.h > --- usr.sbin/vmd/virtio.h > +++ usr.sbin/vmd/virtio.h > @@ -38,7 +38,7 @@ > > /* VMM Control Interface shutdown timeout (in seconds) */ > #define VMMCI_TIMEOUT 3 > -#define VMMCI_SHUTDOWN_TIMEOUT 30 > +#define VMMCI_SHUTDOWN_TIMEOUT 120 > > /* All the devices we support have either 1, 2 or 3 queues */ > /* viornd - 1 queue >
I took a look through the code. I'd say this bump is fine, there is no side effect aside from just waiting for the VM to shutdown, *except* possibly when waiting for the host to shutdown (/etc/rc in the shutdown path), that might take longer if some VMs get stuck in their shutdown code. But if you got impatient, you could always ^C at that point... ok mlarkin -ml
