Hi,
The current behavior when the administrator runs 'ifdown <ovsbr>'
is that the bridge is always destroyed.
Fedora ifdown script:
rhel/etc_sysconfig_network-scripts_ifdown-ovs
[...]
48 case "$TYPE" in
49 OVSBridge)
50 ${OTHERSCRIPT} ${CONFIG} $2
51 retval=$?
*52 ovs-vsctl -t ${TIMEOUT} -- --if-exists del-br "$DEVICE"
53 ;;
Debian ifdown script:
debian/ifupdown.sh
[...]
79 elif [ "${MODE}" = "stop" ]; then
80 case "${IF_OVS_TYPE}" in
81 OVSBridge)
82 if [ ! -z "${IF_OVS_PORTS}" ]; then
83 ifdown --allow="${IFACE}" ${IF_OVS_PORTS}
84 fi
85
*86 ovs_vsctl -- --if-exists del-br "${IFACE}"
87 ;;
Therefore it destroys all the configuration in the DB, the flows, etc...
Linux Bonding, for instance, removes the port when the
'ifdown <bonding port>' is executed but doesn't destroy
the master device.
Linux bridge destroys the bridge only if there is no bridge port
attached, otherwise it remains on the system. Maybe this behavior
is the source of the bug report I got.
On another hand, destroying the devices on 'ifdown' cleans the DB
for a fresh restart.
I can see pros and cons of leaving this as is or changing to just
put the bridge down.
I wonder if anyone else has experienced any other problem or limitation
because of the current behavior.
Thanks,
fbl
_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss
