Otherwise failed non-blocking connection could be reported as
connected. This causes errors in all following operations with the
socket.
At least this is true on FreeBSD, where POLLHUP could be set without
POLLERR.
For example, stream_open_block() tests fails with the following error
reporting successful connection to the 'WRONG_PORT':
./ovsdb-idl.at:1817:
$PYTHON2 $srcdir/test-stream.py tcp:127.0.0.1:$WRONG_PORT
stdout:
./ovsdb-idl.at:1817: exit code was 0, expected 1
2399. ovsdb-idl.at:1817: FAILED (ovsdb-idl.at:1817)
Also added new tests to track this issue in C library:
'Check Stream open block - C - tcp'
'Check Stream open block - C - tcp6'
CC: Numan Siddique <nusiddiq@redhat.com>
Fixes: c1aa16d191d2 ("ovs python: ovs.stream.open_block() returns success even if the remote is unreachable")
Fixes: d6cedfd9d29d ("socket-util: Avoid using SO_ERROR.")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
- Rely on layers l3Offset field to get offset of IP header.
- Aslo fix passing 'newNbl' to IP fragment which is not required.
- Fixed including a header file twice.
Signed-off-by: Anand Kumar <kumaranand@vmware.com>
Acked-by: Alin Gabriel Serdean <aserdean@ovn.org>
Signed-off-by: Alin Gabriel Serdean <aserdean@ovn.org>
sflow.log reports the first ip address of the 'loopback' interface.
It could be different on different systems. For example, on FreeBSD
IPv6 [::1] address goes first despite of IPv4 127.0.0.1 on Linux.
Let's just replace it to IPv4 always to make tests work.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
':a', 'ba' and other suff is not portable.
Additionally removed CORE_QUEUE* macroses as they only complicates
the tests. 'sort' used instead.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
While configuring sFlow agent OVS tries to treat the value as a name
of the interface at first, after that it tries to treat it as an ip
address. While trying to create netdev from the 'agent', netdev-bsd
calls 'netdev_get_flags()' which produces following warning:
failed to get flags for network device 127.0.0.1
This does not happen with netdev-linux because it uses its own
implementation of 'get_flags' while creating the netdev.
Let's just ignore the warning for sFlow tests.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
'read' requires explicit argument.
'sed' partially replaced with more portable 'tr' because '\n'
could not be recognized as a line break.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
'sock.setsockopt' could throw exceptions. For example, if non-blocking
connection failed before the call:
Traceback (most recent call last):
File "../.././test-ovsdb.py", line 896, in <module>
main(sys.argv)
File "../.././test-ovsdb.py", line 891, in main
func(*args)
File "../.././test-ovsdb.py", line 604, in do_idl
ovs.stream.Stream.open(r))
File "/root/git_/ovs/python/ovs/stream.py", line 190, in open
error, sock = cls._open(suffix, dscp)
File "/root/git_/ovs/python/ovs/stream.py", line 744, in _open
sock.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
File "/usr/local/lib/python2.7/socket.py", line 228, in meth
return getattr(self._sock,name)(*args)
socket.error: [Errno 54] Connection reset by peer
This fixes tests on FreeBSD.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
'sed -i' on FreeBSD always expects backup filename extention
passed while GNU version expects it only if specified without
extra space after the '-i'. Let's specify the backup extention
to make BSD sed work.
This fixes test on FreeBSD.
CC: Joe Stringer <joe@ovn.org>
Fixes: 07659514c3c1 ("Add support for connection tracking.")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
strtoull() doesn't necessarily set errno if it finds nothing to parse, but
this code didn't check for that case.
Reported-by: Ilya Maximets <i.maximets@samsung.com>
Reported-at: https://mail.openvswitch.org/pipermail/ovs-dev/2018-December/354622.html
Signed-off-by: Ben Pfaff <blp@ovn.org>
Despite of linux, 'strtoull' on FreeBSD sets errno to EINVAL in case
of no digits found. This causes odp_flow parsing failure if
there is no geneve data inside it. For example, ovs fails to parse
following flow on FreeBSD:
tunnel(<...>,geneve({class=0xffff,type=1,len=0}),<...>)
Moving the parsing attempt under the if condition fixes the following
unit test failure:
tunnel.at:780: testing tunnel - Geneve option present ...
./tunnel.at:810: ovs-appctl ofproto/trace ovs-dummy \
'tunnel(<...>,geneve({class=0xffff,type=1,len=0}), ...'
--- /dev/null 2018-12-18 13:24:55.001110000 +0000
+++ /tmp/cirrus-ci-build/tests/testsuite.dir/at-groups/848/stderr
@@ -0,0 +1,2 @@
+ovs-dummy: unknown bridge
+ovs-appctl: ovs-vswitchd: server returned an error
CC: Jesse Gross <jesse@kernel.org>
Fixes: 622a0a8e764d ("odp-util: Geneve netlink decoding.")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
This keyword is not portable and also optional in bash.
Fixes test on FreeBSD.
CC: Miguel Angel Ajo <majopela@redhat.com>
Fixes: 508b7f961bd6 ("ovn: l3ha, make is_chassis_active aware of gateway_chassis")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Current version is broken because it converts first argument to
integer and after that substracts the duoble value. At the end
the result converted to integer again.
This does not cause unit test failures because qsort on linux
accidentially makes right order. On FreeBSD this leads to the
test failure:
TEST '10-intervals-linear-growth'
Assertion \
'|(&stats)->pctl_95 - (&d->expected_stats)->pctl_95| < 1e-1' failed:
|9 - 10| < 0.1
CC: Mark Michelson <mmichels@redhat.com>
Acked-by: Mark Michelson <mmichels@redhat.com>
Fixes: aed45befeff2 ("Add stopwatch timing API")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Almost equal 'check_dpflow_stats()' functions defined 4 times.
Make it common to reduce code duplication.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Fixes select group test on FreeBSD:
test-source: 27: Syntax error: word unexpected (expecting ")")
CC: Jan Scheurich <jan.scheurich@ericsson.com>
Fixes: 06db81ccfe6d ("ofproto-dpif: Use dp_hash as default selection method")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
'read -d' is a bash extention. Replace it with simple 'cat'.
This fixes 'select group' tests on FreeBSD.
CC: Jan Scheurich <jan.scheurich@ericsson.com>
Fixes: 06db81ccfe6d ("ofproto-dpif: Use dp_hash as default selection method")
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
valgrind tests always runs with '-d' option which breaks the
re-checking. Lets just drop the rechecking support for these
targets.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
This fixes 'RECHECK=yes' functionality in case of configured test
range. For example, before the patch following cmdline will result in
running all of the 1000 tests re-checking in case of any failure:
make -j8 check TESTSUITEFLAGS='1000-1999' RECHECK=yes
This happens because ranges and pattern matching options has higher
priority than the '--recheck'.
With patch, only failed tests from the range will be re-checked.
With this patch applied we're dropping support of '--verbose' and
'--trace' options while re-check, but, IMHO, these options makes
sense mostly while debugging individual tests and not much usable
while running recheck of a whole testsuite.
'--jobs' we're resetting each time anyway.
Not sure if someone overrides default '--directory'. For me it looks
not very useful. Changing the color mode also looks not much
profitable.
OTOH, re-checking the ranges or keyword matched tests is very useful,
for example, if you're trying to split up single testsuite check in
a few independent CI jobs.
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Add check to validate that 'conn_clean()' is only called for
conntrack entries of default 'conn_type'.
Signed-off-by: Darrell Ball <dlu998@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
The user should only reference a conntrack entry by the forward
direction context, as per 'conntrack_flush()', enforce this by
checking for 'default' conn_type. The likelihood of a user
not using the original tuple is low, but it should be guarded
against, logged and documented.
Signed-off-by: Darrell Ball <dlu998@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
When fallback to ephemeral ports triggers to find a NAT translation,
it may happen that the full address range is not explored; i.e. if
all ephemeral ports are being used for the address range >= the
first address checked and there are other addresses in the
available range, then they would not be explored for availability.
The likelihood of hitting this condition is rare. The fix is to
reset the first address to the minimum address when starting to
search ephemeral ports. Found by inspection.
Fixes: 286de2729955 ("dpdk: Userspace Datapath: Introduce NAT Support.")
Signed-off-by: Darrell Ball <dlu998@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Ephemeral port fallback is being done for DNAT and the code could be hit in
some special cases and testing configurations. Also good packets are
expected to be persistently dropped in this case, which is not a common
user goal. Regardless, this is incorrect, so filter this out. Also, rename
the variable used for checking whether ephemeral ports need to be checked.
Reported-at: https://mail.openvswitch.org/pipermail/ovs-dev/2018-August/351629.html
Fixes: 286de2729955 ("dpdk: Userspace Datapath: Introduce NAT Support.")
Signed-off-by: Darrell Ball <dlu998@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
This reverts commit 2f748bf8016c ("datapath: Derive IP protocol...")
This commit is causing some ipv6 fragmentation errors in some older
kernels. Revert for now and then we can determine how to implement
this patch with appropriate compatability layer changes to prevent
errors on older kernels.
CC: Yi-Hung Wei <yihung.wei@gmail.com>
Signed-off-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
The original static key backport is based on the upstream
net-next commit 11276d5306b8
("locking/static_keys: Add a new static_key interface").
However, in Canonical's Trusty kernel, it introduced partial static
support which have different definition of some of the macros that
breaks the compatibility code.
For example, in net-next git tree commit 11276d5306b8
("locking/static_keys: Add a new static_key interface").
+ #define DEFINE_STATIC_KEY_TRUE(name) \
+ struct static_key_true name = STATIC_KEY_TRUE_INIT
On the other hand, in Canonical's Trusty git tree commit 13f5d5d1cccb6
("x86/KVM/VMX: Add module argument for L1TF mitigation")
+ #define DEFINE_STATIC_KEY_TRUE(name) \
+ struct static_key name = STATIC_KEY_INIT_TRUE
This commit resolves the ovs kernel module compatibility issue on
Trusty kernel.
VMware-BZ: #2251101
Fixes: 6660a9597a49 ("datapath: compat: Introduce static key support")
Tested-by: Greg Rose <gvrose8192@gmail.com>
Reviewed-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Yi-Hung Wei <yihung.wei@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
When an IPv6 packet enters a router pipeline and it needs to be routed via
the nexthop IP address set in the static route, OVN generates an IPv6
Neigh Solicitation request if the nexthop IP is not resolved yet. But
right now, the generated IPv6 Neigh Solicitation packet doesn't set
the eth.dst to the mutlicast address derived from the nexthop and
ip6.dst to the solicited-node multicast address corresponding to the
nexthop address. Instead it generates these values from the actual
ip6.dst of the original packet.
This patch fixes this issue.
Signed-off-by: Numan Siddique <nusiddiq@redhat.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Upstream commit:
commit b233504033dbd65740e59681820ccfd0a2a8ec53
Author: Yifeng Sun <pkusunyifeng@gmail.com>
Date: Mon Jul 2 08:18:03 2018 -0700
openvswitch: kernel datapath clone action
Add 'clone' action to kernel datapath by using existing functions.
When actions within clone don't modify the current flow, the flow
key is not cloned before executing clone actions.
This is a follow up patch for this incomplete work:
https://patchwork.ozlabs.org/patch/722096/
v1 -> v2:
Refactor as advised by reviewer.
Signed-off-by: Yifeng Sun <pkusunyifeng@gmail.com>
Signed-off-by: Andy Zhou <azhou@ovn.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Tested-by: Greg Rose <gvrose8192@gmail.com>
Reviewed-by: Greg Rose <gvrose8192@gmail.com>
Co-authored-by: Andy Zhou <azhou@ovn.org>
Signed-off-by: Yifeng Sun <pkusunyifeng@gmail.com>
Signed-off-by: Andy Zhou <azhou@ovn.org>
Signed-off-by: Ben Pfaff <blp@ovn.org>
DPDK 18.11 uses libmnl when MLX5 PMD is enabled.
This commit makes OVS to link to libmnl when MLX5 PMD is enabled on
DPDK.
Signed-off-by: Timothy Redaelli <tredaelli@redhat.com>
Signed-off-by: Ian Stokes <ian.stokes@intel.com>
When libtool is in the picture to deal with shared libraries, processes
end up with an "lt-" prefix in their process names. This caused the
process name check in daemon.at to fail. This commit fixes the problem by
stripping off that prefix.
Fixes: d8c6955a03ea ("tests: Simplify and improve the daemon tests.")
Reported-by: Timothy Redaelli <tredaelli@redhat.com>
Reported-at: https://mail.openvswitch.org/pipermail/ovs-dev/2018-December/354574.html
Acked-by: Timothy Redaelli <tredaelli@redhat.com>
Tested-by: Timothy Redaelli <tredaelli@redhat.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Upstream commit:
commit 43d0e96022ae3c66743c01bba6c18a3afec7b578
Author: Colin Ian King <colin.king@canonical.com>
Date: Tue Nov 27 14:37:17 2018 +0000
openvswitch: fix spelling mistake "execeeds" -> "exceeds"
There is a spelling mistake in a net_warn_ratelimited message, fix this.
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Simon Horman <simon.horman@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
CC: Colin Ian King <colin.king@canonical.com>
Acked-by: William Tu <u9012063@gmail.com>
Signed-off-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Upstream commit:
commit eddf11e18dff0e8671e06ce54e64cfc843303ab9
Author: YueHaibing <yuehaibing@huawei.com>
Date: Wed Sep 26 17:15:38 2018 +0800
net: ovs: fix return type of ndo_start_xmit function
The method ndo_start_xmit() is defined as returning an 'netdev_tx_t',
which is a typedef for an enum type, so make sure the implementation in
this driver has returns 'netdev_tx_t' value, and change the function
return type to netdev_tx_t.
Found by coccinelle.
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
CC: YueHaibing <yuehaibing@huawei.com>
Acked-by: William Tu <u9012063@gmail.com>
Signed-off-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Upstream commit:
commit fa642f08839bf2ff35b2f6c6a6c062aee8121ba8
Author: Yi-Hung Wei <yihung.wei@gmail.com>
Date: Tue Sep 4 15:33:41 2018 -0700
openvswitch: Derive IP protocol number for IPv6 later frags
Currently, OVS only parses the IP protocol number for the first
IPv6 fragment, but sets the IP protocol number for the later fragments
to be NEXTHDF_FRAGMENT. This patch tries to derive the IP protocol
number for the IPV6 later frags so that we can match that.
Signed-off-by: Yi-Hung Wei <yihung.wei@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
CC: Yi-Hung Wei <yihung.wei@gmail.com>
Acked-by: Yi-Hung Wei <yihung.wei@gmail.com>
Signed-off-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Upstream commit:
commit 256c87c17c53e60882a43dcf3e98f3bf859eaf6f
Author: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Date: Tue Jun 26 21:39:36 2018 -0700
net: check tunnel option type in tunnel flags
Check the tunnel option type stored in tunnel flags when creating options
for tunnels. Thereby ensuring we do not set geneve, vxlan or erspan tunnel
options on interfaces that are not associated with them.
Make sure all users of the infrastructure set correct flags, for the BPF
helper we have to set all bits to keep backward compatibility.
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
CC: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Acked-by: William Tu <u9012063@gmail.com>
Signed-off-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
A new flags parameter has been added in 4.19 so add compat fixup.
Reviewed-by: Yifeng Sun <pkusunyifeng@gmail.com>
Signed-off-by: Greg Rose <gvrose8192@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
This commit adds support for DPDK v18.11, it includes the following
changes.
1. Enable compilation and linkage with dpdk 18.11.0
The following dpdk commits which were introduced after dpdk 17.11.x
require OVS updates to accommodate to the dpdk changes.
- ce17edde ("ethdev: introduce Rx queue offloads API")
- ab3ce1e0 ("ethdev: remove old offload API")
- c06ddf96 ("meter: add configuration profile")
- e58638c3 ("ethdev: fix TPID handling in flow API")
- cd8c7c7c ("ethdev: replace bus specific struct with generic dev")
- ac8d22de ("ethdev: flatten RSS configuration in flow API")
2. Limit configured rss hash functions to only those supported
by the eth device.
3. Set default RSS key in struct action_rss_data, required by OVS
commit- e8a2b5bf ("netdev-dpdk: implement flow offload with rte flow")
when configured with "other_config:hw-offload=true".
4. DEV_RX_OFFLOAD_CRC_STRIP has been removed from DPDK 18.11.
DEV_RX_OFFLOAD_KEEP_CRC can now be used to keep the CRC.
Use the correct flag and check it is supported.
5. rte_eth_dev_attach/detach have been removed from DPDK 18.11.
Replace them with rte_dev_probe/remove.
6. Update docs and travis to use DPDK18.11.
This commit squashes the following commits present on the dpdk-latest
branch:
7f021f902bb3 ("netdev-dpdk: Upgrade to dpdk v18.08")
270d9216f1ed ("netdev-dpdk: Set scatter based on capabilities")
bef2cdc8f412 ("netdev-dpdk: Fix returning the field of malloced struct.")
73c1a65167fc ("redhat: change variable used for non-root user support")
eb485f60ce44 ("dpdk: Update to use DPDK 18.11.")
For credit all authors of the original commits above have been added as
co-authors for this commmit.
From: Ophir Munk <ophirmu@mellanox.com>
Signed-off-by: Ophir Munk <ophirmu@mellanox.com>
Signed-off-by: Kevin Traynor <ktraynor@redhat.com>
Co-authored-by: Kevin Traynor <ktraynor@redhat.com>
Signed-off-by: Ilya Maximets <i.maximets@samsung.com>
Co-authored-by: Ilya Maximets <i.maximets@samsung.com>
Signed-off-by: Timothy Redaelli <tredaelli@redhat.com>
Co-authored-by: Timothy Redaelli <tredaelli@redhat.com>
Signed-off-by: Ian Stokes <ian.stokes@intel.com>
Add selected dynamic mac address to MACAM in update_dynamic_addresses
and not just in in ipam_add_port_addresses/ipam_insert_lsp_addresses
since the second approach can produce a duplicated L2 address in a
IPv6-only network if ipv6_prefix is provided after logical port creation.
The issue can be triggered with the following reproducer:
$ovn-nbctl ls-add sw0
$ovn-nbctl lsp-add sw0 sw0-port1
$ovn-nbctl lsp-set-addresses sw0-port1 "dynamic"
$ovn-nbctl lsp-add sw0 sw0-port2
$ovn-nbctl lsp-set-addresses sw0-port2 "dynamic"
$ovs-vsctl add-port br-int p1 -- \
set Interface p1 external_ids:iface-id=sw0-port1
$ovs-vsctl add-port br-int p2 -- \
set Interface p2 external_ids:iface-id=sw0-port2
[..]
$ovn-nbctl --wait=sb set Logical-switch sw0 \
other_config:ipv6_prefix="aef0::"
$ovn-nbctl list logical_switch_port
_uuid : 1e0e2ed8-20c6-48dc-bfa8-d823e48c9f45
addresses : [dynamic]
dhcpv4_options : []
dhcpv6_options : []
dynamic_addresses : "0a:00:00:00:00:01 aef0::800:ff:fe00:1"
enabled : []
external_ids : {}
name : "sw0-port1"
options : {}
parent_name : []
port_security : []
tag : []
tag_request : []
type : ""
up : true
_uuid : cfeab7fb-e20b-41f1-974c-f99e0b5293d7
addresses : [dynamic]
dhcpv4_options : []
dhcpv6_options : []
dynamic_addresses : "0a:00:00:00:00:01 aef0::800:ff:fe00:1"
enabled : []
external_ids : {}
name : "sw0-port2"
options : {}
parent_name : []
port_security : []
tag : []
tag_request : []
type : ""
up : true
Fixes: c814545b43ac ("OVN: configure L2 address according to the used IP
address")
Acked-by: Mark Michelson <mmichels@redhat.com>
Acked-by: Numan Siddique <nusiddiq@redhat.com>
Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi@redhat.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
See RFC 2131, section 4.3.2. When handling a DHCPREQUEST message, the
server should validate that the client's requested IP matches the
offered IP. If not, the server should reply with a DHCPNAK. The client's
requested IP is either specified as the Requested IP Address (option
50), or as the ciaddr, depending on the client's state.
Signed-off-by: Gregory Smith <gasmith@nutanix.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
This structure isn't performance-sensitive and making it packed simplifies
thinking about access to it.
CC: Gregory Smith <gasmith@nutanix.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
Prior to this patch, GARPs announcing NAT addresses or new VIFs
were sent out to localnet ofport through an output action.
This can lead to problems since local datapaths won't get those
GARPs and ovn-controller won't update MAC_Binding entries (as
upstream switch will not send back the GARP to this port hence
other logical routers won't update their neighbours).
This patch is changing the behavior so that GARPs get injected
to OVN pipeline of the external switch. This way, they'll get
broadcasted to local pipelines and also sent out to the external
network through the localnet port.
Acked-by: Han Zhou <hzhou8@ebay.com>
Acked-by: Numan Siddique <nusiddiq@redhat.com>
Reported-at: https://mail.openvswitch.org/pipermail/ovs-discuss/2018-October/047604.html
Signed-off-by: Daniel Alvarez <dalvarez@redhat.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
This patch [0] in glibc fixes an issue which is right now workarounded
in OVS by [1]. I'm adding a comment to indicate that from glibc 2.28
and beyond, the workaround is not needed so that we can eventually
remove it.
[0] https://sourceware.org/git/gitweb.cgi?p=glibc.git;h=c1f86a33ca32e26a9d6e29fc961e5ecb5e2e5eb4
[1] 3434d30686
Signed-off-by: Daniel Alvarez <dalvarez@redhat.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
