Re: [PATCH net] failover: eliminate callback hell

From
Siwei Liu <hidden>
Date
2018-06-11 18:56:58
Reply to
Jakub Kicinski, 2018-06-09
On Fri, Jun 8, 2018 at 6:29 PM, Jakub Kicinski [off-list ref] wrote:
On Fri, 8 Jun 2018 16:44:12 -0700, Siwei Liu wrote:
quoted
quoted
quoted
I have a somewhat different view regarding IFF_HIDDEN. The purpose of
that flag, as well as the 1-netdev model, is to have a means to
inherit the interface name from the VF, and to eliminate playing hacks
around renaming devices, customizing udev rules and et al. Why
inheriting VF's name important? To allow existing config/setup around
VF continues to work across kernel feature upgrade. Most of network
config files in all distros are based on interface names. Few are MAC
address based but making lower slaves hidden would cover the rest. And
most importantly, preserving the same level of user experience as
using raw VF interface once getting all ndo_ops and ethtool_ops
exposed. This is essential to realize transparent live migration that
users dont have to learn and be aware of the undertaken.
Inheriting the VF name will fail in the migration scenario.
It is perfectly reasonable to migrate a guest to another machine where
the VF PCI address is different. And since current udev/systemd model
is to base network device name off of PCI address, the device will change
name when guest is migrated.
The scenario of having VF on a different PCI address on post migration
is essentially equal to plugging in a new NIC. Why it has to pair with
the original PV? A sepearte PV device should be in place to pair the
new VF.
IMHO it may be a better idea to look at the VF as acceleration for the
PV rather than PV a migration vehicle from the VF.  Hence we should
I'm basically talking about two use cases not about solutions or
implementations specifically. As said, the one I'm looking into needs
to migrate a pre-failover VF setup to 1-netdev failover model in a
transparent manner. There's no point to switch PCI address back and
forth in the backend to set where to bind the PV or the VF, as you
have no ways to predict what guest kernel will be running until its
fully loaded. Supporting a VF on new location binding to existing PV
might be nice, but not directly relevant to those who don't need this
side feature than migration itself.

Having said that, while I somewhat agree both use cases should have
its own place in the picture, I don't think judging one better than
the other or vice versa is logical IMHO.
continue to follow the naming of PV, like the current implementation
does implicitly by linking to PV's struct device.
The current implementation may only work with new userspace, even so
the eth0/eth0nsby naming is not consistenly persisted due to races in
bus probing. The naming part should be fixed.

-Siwei
Thread (52 messages) 52 messages, 8 authors, last reply 2018-06-12
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help