Date: Sat, 10 Dec 2016 23:20:28 -0800 From: Peter Grehan <grehan@freebsd.org> To: Rajil Saraswat <rajil.s@gmail.com> Cc: freebsd-virtualization@FreeBSD.org Subject: Re: Debian 8 CPU stall Message-ID: <56863fec-d8f8-2601-811f-511cd199e82f@freebsd.org> In-Reply-To: <6cff5bf2-9654-8627-83c4-6ab48ee763b5@gmail.com> References: <b011b080-5637-2da5-8a8a-819b7b1fabd3@gmail.com> <9c9e83a5-16c6-0ab5-0ac4-af0a54430706@freebsd.org> <6cff5bf2-9654-8627-83c4-6ab48ee763b5@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Rajil, > No the system is not oversubscribed. I have a 11 vCPU (1 on debian and 1 > on ubuntu) on a 24 core machine. The debian jail is running x2go and an > ssh server for remote access, so the I/O shouldnt be an issue. The > ubuntu jail doesnt give out any warning messages though. ... > I use vm-bhyve for managing the jails which looks like this for debian8: > > guest="linux" > loader="grub" > cpu=1 > memory=2048M > network0_type="virtio-net" > network0_switch="lannetwork" > disk0_type="virtio-blk" > disk0_name="/dev/zvol/vmpool/os2" > disk0_dev="custom" > passthru0="2/0/0" > passthru1="2/0/1" Looks like a possibly lost interrupt. Can you check to see if this system is using APIC virtualization ? # sysctl -a hw.vmm.vmx.cap ... hw.vmm.vmx.cap.posted_interrupts: 1 hw.vmm.vmx.cap.virtual_interrupt_delivery: 1 (these will be zero if APIC virtualization not in use) If so, one thing to try is disabling these and see if that helps. This can be done by setting the tunable hw.vmm.vmx.use_apic_vid to 1 (e.g. in /boot/loader.conf, or unloading vmm.ko, setting the value with kenv, and the reloading vmm.ko) later, Peter.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?56863fec-d8f8-2601-811f-511cd199e82f>