On Tue, Feb 02, 2016 at 06:28:58PM +0900, Hajimu UMEMOTO
wrote:> Hi,
>
> >>>>> On Tue, 2 Feb 2016 08:54:17 +0000
> >>>>> Steven Hartland <killing at multiplay.co.uk>
said:
>
> killing> Some more information about your enviroment would be helpful:
> killing> 1. What revision of stable/10 are you running?
>
> It occurs on r292895 and later.
>
> killing> 2. What workloads are you running?
>
> Apache, Cyrus IMAPd, BIND and mpd5. However, traffic is almost
> nothing without running the periodic daily scripts.
>
> killing> 3. What's the output of procstat -k -k when this happens
(assuming its
> killing> possible to run)?
>
> I could not get it.
>
> killing> 4. What's the output of sysctl -a |grep vnode, usually and
when this
> killing> happens?
>
> The following is usual case:
>
> kern.maxvnodes: 400000
> kern.minvnodes: 100000
> Syncing disks, vnodes remaining...0 0 0 0 0 done
> Syncing disks, vnodes remaining...0 0 0 0 0 0 done
> vm.stats.vm.v_vnodepgsout: 129
> vm.stats.vm.v_vnodepgsin: 836656
> vm.stats.vm.v_vnodeout: 101
> vm.stats.vm.v_vnodein: 90562
> vfs.freevnodes: 39680
> vfs.wantfreevnodes: 100000
> vfs.vnodes_created: 141735
> vfs.numvnodes: 59118
> debug.sizeof.vnode: 472
Please gather the information listed at
https://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kerneldebug-deadlocks.html