On 8/13/25 8:05 PM, Aaron Conole via dev wrote: > When adjusting bond parameters, any adjustment is considered sufficient > for triggering a rebalance. This is a very simplistic config update > scheme that triggers a complete rebalance even if the time adjustment > would move the next expiration out beyond the last calculated expiration. > > For the interval parameter only, we can simply recalculate the expiry > deadline and let the next bond_run() event do the rebalance if needed. > Even if the recalculation would cause the deadline to have occurred in > the past, it should execute on the next bond_run() anyway. This is > still okay, as the rebalance interval timeout may not result in a > full rebalance anyway. > > Reported-at: > https://www.mail-archive.com/[email protected]/msg10409.html
This should be a link to the primary ovs-discuss archives: https://mail.openvswitch.org/pipermail/ovs-discuss/2025-April/053588.html > Signed-off-by: Aaron Conole <[email protected]> > --- > ofproto/bond.c | 17 ++++++++++++++++- > 1 file changed, 16 insertions(+), 1 deletion(-) > > diff --git a/ofproto/bond.c b/ofproto/bond.c > index 3859ddca08..00f5dc754c 100644 > --- a/ofproto/bond.c > +++ b/ofproto/bond.c > @@ -459,8 +459,23 @@ bond_reconfigure(struct bond *bond, const struct > bond_settings *s) > } > > if (bond->rebalance_interval != s->rebalance_interval) { > + if (s->rebalance_interval && bond->rebalance_interval) { > + /* Recompute the next rebalance interval by moving the > + * next_rebalance to be offset by the new interval. In this > + * case, if all that was updated is the rebalance interval, > + * we can skip triggering the rest of the port reconfiguration. > */ > + if (bond->next_rebalance) { > + long long int old_start_time = > + bond->next_rebalance - s->rebalance_interval; > + bond->next_rebalance = > + old_start_time + bond->rebalance_interval; > + } > + } else { > + /* When the bond is doing a disable/enable of the rebalance > + * interval, trigger the revalidation. */ > + revalidate = true; > + } > bond->rebalance_interval = s->rebalance_interval; > - revalidate = true; > } > > if (bond->balance != s->balance) { Would be nice to have a test for this, but we can add one while fixing the rest of the issues discussed in v1. With the link update (may be fixed while applying, I suppose): Acked-by: Ilya Maximets <[email protected]> _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
