11.2-RELEASE panics with a bit of load
meowthink at gmail.com
Thu Oct 4 10:57:44 UTC 2018
I do noticed that the default vfs.zfs.arc_max seems too aggressive, as
it will cause tings like rsync swapping in and out, esp. on configs
having 32GB RAM, and smaller.
So if a reduced arc_max works, I'd suggest you check your swap device.
Some SSDs are known to have issues under huge IO workload, like high
latency thus abort host commands and force the system reties on read.
On Thu, Oct 4, 2018 at 3:06 PM <rainer at ultra-secure.de> wrote:
> Am 2018-10-04 05:53, schrieb Holger Kipp:
> > Dear Rainer,
> > if no one else is experiencing these errors, then it is probably a
> > hardware error.
> > I had once encountered problems with incompatible memory (combination
> > of memory (ok with other boards) and board (ok with other memory) did
> > not work properly). The issue could be triggered with a simple
> > buildworld within minutes. However memtest86 could run for hours
> > without finding a thing.
> > Please don‘t rely on memtest86 if it does not report any problems.
> > To be sure: check / adjust memory timings, remove and/or replace
> > memory modules and try to trigger the error again.
> > Best regards,
> > Holger Kipp
> I have this problem on two machines, actually.
> It would be highly unlikely for both to have hardware problems.
> Though possible, of course.
> Reducing ARC seems to help, though, and I assume that is the problem the
> machine ran into because I still have 29GB out of 32GB wired now, with
> an ARC limited to 60% of kmem_size.
> freebsd-stable at freebsd.org mailing list
> To unsubscribe, send any mail to "freebsd-stable-unsubscribe at freebsd.org"
More information about the freebsd-stable