Date: Wed, 19 Dec 2018 19:36:29 +0000 From: bugzilla-noreply@freebsd.org To: virtualization@FreeBSD.org Subject: [Bug 231117] I/O lockups inside bhyve vms Message-ID: <bug-231117-27103-mACk8raKti@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-231117-27103@https.bugs.freebsd.org/bugzilla/> References: <bug-231117-27103@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D231117 --- Comment #11 from Kristian K. Christensen <kkc@duckpond.dk> --- (In reply to Allan Jude from comment #10) I'll give it a shot, but expect it to take a little while to get the output. Let me quickly explain: Currently, the only thing that stops doing io is our bhyve instances running anything 11.2 or newer. (11.1 rarely ever have issues, though it *does* hap= pen, see Niels' comment on how it differs). However, it's the entire instance th= at stops doing io. The zvol simply receives no updates from here on out. Regretably, on 11.2+, it's mostly on production vms we can effectively reproduce it often enough to debug on it, so we tend to roll them back after testing. I can reproduce on test vms, but they crash more infrequently and = thus it could be a long wait. Furthermore, ALL writes stop and any process that wants to write to anything, stalls. This easily gives me a pid to debug wit= h, but very little access to run commands on the vm. The only fix is to kill t= he bhyve pid and start it up again. We can run 11.2+ on the hypervisor, it's o= nly running 11.2+ inside the VM that causes the hang. VMs running UFS do not ha= ng. I'm going to leave a couple of sessions logged in and run procstat and ps fairly often, to keep them in memory, but I suspect it can take me a little= bit to get the required output. I'll get back to you, I'm going to assume zfskern is the most interesting p= id to get output from, but if any random that can't write will do, let me know. --=20 You are receiving this mail because: You are the assignee for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-231117-27103-mACk8raKti>