Date: Sat, 29 Nov 2025 15:20:05 +0100 From: Roman Bogorodskiy <novel@freebsd.org> To: virtualization@freebsd.org Subject: bhyve(8) suspend interface Message-ID: <aSsBFSalrvpY1kxm@tulp>
next in thread | raw e-mail | index | archive | help
Hi, While experimenting with the bhyve suspend/resume feature I noticed a couple of things which seem a little confusing. The first thing is bhyve(8) exit code after suspend. If I run: bhyve -c 2 -m 4096 ... my_vm And then run: bhyvectl --suspend=/var/run/vms/my_vm --vm my_vm Once suspend completes, the bhyve process exits with retcode 0. According to the bhyve(8) manual page, this exit code corresponds to a VM reboot. It also seems inconvenient from management software/scripting point of view, because reaction to this exit code is to boot a VM again (assuming that if a user initiated a reboot, they expect their VM to boot again soon). My impression is that it would be more natural to group it with other events like 1:power off or 2:halted, which do not imply immediate VM restart. Or add a dedicated exit code, say, 5:suspended? The second question is that, apparently, calling bhyvectl --suspend ... is not enough, as after suspend completion the VM entity is still present in /dev/vmm, so it looks like bhyvectl --destroy is necessary. It's not obvious from the bhyvectl(8) manual page and generally it's not obvious why two bhyvectl(8) commands are necessary to suspend a VM. Interestingly, bhyvectl(8) does not complain early about using "--suspend" and "--destroy" at the same time, i.e.: bhyvectl --suspend=/var/run/vms/my_vm --destroy --vm my_vm But fails with obscure messages: bhyvectl(8): connect() failed: Connection refused bhyve(8): vm_run error -1, errno 6 Failed to emulate instruction sequence [ 8b800040000089c048018108060000 ] at 0xffffffffc06d85f3 And dumps a core. Roman
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?aSsBFSalrvpY1kxm>