I'm going to test setting this parameter (although I don't use Xen):
http://www.drbd.org/users-guide/s-xen-drbd-mod-params.html
Googling around there seems to be a relation between Oops, drbd
active/active and this setting.
DRBD source code says this which sounds like it could be related:
*/* e.g. XFS meta- & log-data is in slab pages, which have a*
* * page_count of 0 and/or have PageSlab() set.*
* * we cannot use send_page for those, as that does get_page();*
* * put_page(); and would cause either a VM_BUG directly, or*
* * __page_cache_release a page that would actually still be
referenced*
* * by someone, leading to some obscure delayed Oops
somewhere else. */*
if (disable_sendpage
<http://lxr.free-electrons.com/ident?i=disable_sendpage> ||
(page_count <http://lxr.free-electrons.com/ident?i=page_count>(page
<http://lxr.free-electrons.com/ident?i=page>) < 1) || PageSlab(page
<http://lxr.free-electrons.com/ident?i=page>))
return _drbd_no_send_page
<http://lxr.free-electrons.com/ident?i=_drbd_no_send_page>(mdev, page
<http://lxr.free-electrons.com/ident?i=page>, offset
<http://lxr.free-electrons.com/ident?i=offset>, size
<http://lxr.free-electrons.com/ident?i=size>, msg_flags);
--
"Judge me by my size, do you? And well you should not"
Master Yoda
