We meet a problem that same tcp session selects different ovs group bucket when
in tcp retransmition, and we can easily reproduce this phenomenon.
After some code research, we found that when tcp retransmit, it may call
function sk_rethink_txhash(), and this function makes skb->hash changed, henc
Bleep bloop. Greetings Marcelo Ricardo Leitner, I am a robot and I have tried
out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 80 characters long (recommended limit is 79)
#32 FILE: vswitchd/vsw
From: wenxu
TC flower doesn't support some ct state flags such as
INVALID/SNAT/DNAT/REPLY. So it is better to reject this rule.
Signed-off-by: wenxu
---
lib/netdev-offload-tc.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/lib/netdev-offload-tc.c b/lib/netdev-offload
hw-offload is can actually be enabled without a restart. Then, once
enabled, a restart is needed in order to make the new value effective.
tc-policy, just like hw-offload, is protected by ovsthread_once_start()
in netdev_set_flow_api_enabled() and is only parsed when hw-offload is
enabled. So lets
On Tue, Feb 2, 2021 at 9:32 AM Ilya Maximets wrote:
>
> On 1/21/21 7:51 PM, Yi-Hung Wei wrote:
> > This patch adds an RPMBUILD_OPT so that user can enable
> > AF_XDP support in the rpm package by:
> >
> > $ make rpm-fedora RPMBUILD_OPT="--with afxdp"
>
> Thanks!
> This change looks good to me in g
This patch adds an RPMBUILD_OPT so that user can enable
AF_XDP support in the rpm package by:
$ make rpm-fedora RPMBUILD_OPT="--with afxdp"
Signed-off-by: Yi-Hung Wei
---
Documentation/intro/install/fedora.rst | 7 +++
rhel/openvswitch-fedora.spec.in| 8
2 files changed, 15
Mellanox card has different XSK design. It requires users to create
dedicated queues for XSK. Unlike Intel's NIC which loads XDP program
to all queues, Mellanox only loads XDP program to a subset of its queue.
When OVS uses AF_XDP with mlx5, it doesn't replace the existing RX and TX
queues in the
Bleep bloop. Greetings Marcelo Ricardo Leitner, I am a robot and I have tried
out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 80 characters long (recommended limit is 79)
#31 FILE: vswitchd/vsw
On Fri, Jan 29, 2021 at 03:03:05PM +0100, Ilya Maximets wrote:
> Some manpages are generated from rST, but these are not included
> in 'dist-docs' make target.
>
> Fixes: fd0837a76f4c ("doc: Convert ovs-vlan-test to rST")
> Signed-off-by: Ilya Maximets
Works for me.
Acked-by: Ben Pfaff
___
The supported upgrade procedure is to always upgrade ovn-controllers
first and OVN DBs and ovn-northd later. This leads however to the
situation when ovn-controller might try to set the Port_Binding.up field
while the Southbound DB isn't yet aware of this field which would
trigger transaction fail
In general, ovn-northd expects ovn-controller to set Port_Binding.up
before it declares the logical switch port as being up.
Even though the recommended upgrade procedure for OVN states that
ovn-controllers should be upgraded before ovn-northd, there are cases
when CMSs don't follow this guideline
For port bindings that are children of other port bindings (container
and virtual ports) set Port_Binding.up directly, when claimed, if their
parent bindings are already up.
For non-VIF port bindings maintain compatibility with older versions
and set Port_Binding.up as soon as they are claimed.
R
First two patches of the series fix issues related to:
- setting Port_Binding.up for container/virtual ports/
- correctly dealing with upgrades when ovn-controller is upgraded first
(recommended way).
The third patch of the series tries to handle out of order upgrades
(ovn-northd before ovn-cont
hw-offload is can actually be enabled without a restart. Then, once
enabled, a restart is needed in order to make the new value effective.
tc-policy, just like hw-offload, is protected by ovsthread_once_start()
in netdev_set_flow_api_enabled() and is only parsed when hw-offload is
enabled. So lets
Bleep bloop. Greetings Dumitru Ceara, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
git-am:
error: Failed to merge in the changes.
hint: Use 'git am --show-current-patch' to see the failed p
Currently when setting 'userspace-tso-enable=true', tunnel test cases
fail due to incorrect checksum, at inner header and outer header.
The patch recalculates the checksum before packet is outputting to
a port (tunnel and tap), makes sure the receiver sees correct checksum.
Consider the following
On 2/3/21 3:26 PM, Numan Siddique wrote:
On Fri, Jan 15, 2021 at 11:56 PM Dumitru Ceara wrote:
In case traffic that gets load balanced is DNAT-ed to a backend IP that
happens to be the source of the traffic then OVN performs an additional
SNAT to ensure that return traffic is directed through
On 2/2/21 8:17 PM, Dumitru Ceara wrote:
On 2/2/21 7:18 PM, Numan Siddique wrote:
On Tue, Feb 2, 2021 at 11:26 PM Numan Siddique wrote:
On Tue, Feb 2, 2021 at 11:11 PM Ben Pfaff wrote:
The test "ovn -- nested containers" adds, removes, and then re-adds a
container port bar2. It appears to
In general, ovn-northd expects ovn-controller to set Port_Binding.up
before it declares the logical switch port as being up.
Even though the recommended upgrade procedure for OVN states that
ovn-controllers should be upgraded before ovn-northd, there are cases
when CMSs don't follow this guideline
The supported upgrade procedure is to always upgrade ovn-controllers
first and OVN DBs and ovn-northd later. This leads however to the
situation when ovn-controller might try to set the Port_Binding.up field
while the Southbound DB isn't yet aware of this field which would
trigger transaction fail
For port bindings that are children of other port bindings (container
and virtual ports) set Port_Binding.up directly, when claimed, if their
parent bindings are already up.
For non-VIF port bindings maintain compatibility with older versions
and set Port_Binding.up as soon as they are claimed.
R
First two patches of the series fix issues related to:
- setting Port_Binding.up for container/virtual ports/
- correctly dealing with upgrades when ovn-controller is upgraded first
(recommended way).
The third patch of the series tries to handle out of order upgrades
(ovn-northd before ovn-cont
In the current ovn-nbctl lr-route-add implementation is possible to add
multiple duplicated routes adding --ecmp or --ecmp-symmetric-reply
option, e.g:
$ovn-nbctl --ecmp-symmetric-reply lr-route-add lr0 10.244.0.7/32 172.18.0.4
$ovn-nbctl --ecmp-symmetric-reply lr-route-add lr0 10.244.0.7/32 172.1
On Wed, Feb 3, 2021 at 1:25 AM Ilya Maximets wrote:
>
> On 1/25/21 2:34 PM, Tonghao Zhang wrote:
> > On Fri, Jan 22, 2021 at 10:50 PM Timothy Redaelli
> > wrote:
> >>
> >> With DPDK 20.11, meson and pkgconfig are used instead of the old
> >> Makefile-based system and so --with-dpdk option is cha
From: Tonghao Zhang
Now OvS supports building with dpdk 20.11, and uses
the '--with-dpdk=shared' option in fedora.spec. Then
change version of dpdk 17.05.1 to 20.11.
Note that dpdk-devel-20.11.x is not released in fedora
distro, but use '20.11' is fine.
Fixes: 252e1e576443 ("dpdk: Update to use
On Fri, Jan 15, 2021 at 11:56 PM Dumitru Ceara wrote:
>
> In case traffic that gets load balanced is DNAT-ed to a backend IP that
> happens to be the source of the traffic then OVN performs an additional
> SNAT to ensure that return traffic is directed through OVN.
>
> Until now the load balancer
On 2/3/21 1:39 PM, Ilya Maximets wrote:
> On 2/3/21 12:58 PM, Marcelo Leitner wrote:
>> On Wed, Feb 03, 2021 at 10:54:19AM +0200, Paul Blakey wrote:
>>>
>>>
>>> On Tue, 2 Feb 2021, Marcelo Leitner wrote:
>>>
On Tue, Feb 02, 2021 at 04:59:44PM +0100, Ilya Maximets wrote:
> On 2/2/21 4:52 PM
Thanks!
Regards,
Vladislav Odintsov
On 03.02.2021, 15:34, "dev on behalf of Numan Siddique"
wrote:
On Wed, Feb 3, 2021 at 2:15 PM wrote:
>
> From: Vladislav Odintsov
>
> New ovn-ic.service and ovn-ic-db.service systemd units were added
> to manage ovn-ic and ovsd
On Thu, Jan 28, 2021 at 10:21 PM Dumitru Ceara wrote:
>
> On 1/27/21 3:23 PM, Frode Nordahl wrote:
> > As per [0] the OVN project now has a separate Patchwork Instance,
> > let's reflect that fact in the documentation.
> >
> > 0: https://mail.openvswitch.org/pipermail/ovs-dev/2020-September/374317
On 2/3/21 12:58 PM, Marcelo Leitner wrote:
> On Wed, Feb 03, 2021 at 10:54:19AM +0200, Paul Blakey wrote:
>>
>>
>> On Tue, 2 Feb 2021, Marcelo Leitner wrote:
>>
>>> On Tue, Feb 02, 2021 at 04:59:44PM +0100, Ilya Maximets wrote:
On 2/2/21 4:52 PM, wenxu wrote:
>
> Hi,
>
> just i
On Wed, Feb 3, 2021 at 2:15 PM wrote:
>
> From: Vladislav Odintsov
>
> New ovn-ic.service and ovn-ic-db.service systemd units were added
> to manage ovn-ic and ovsdb-servers for OVN IC respectively.
> ovn-ic-db.service manages both OVN_IC_Northbound and OVN_IC_Southbound
> OVSDB server processes.
On Wed, Feb 3, 2021 at 5:43 PM Dumitru Ceara wrote:
>
> On 2/3/21 12:06 PM, Numan Siddique wrote:
> > On Tue, Feb 2, 2021 at 3:15 PM Dumitru Ceara wrote:
> >>
> >> The supported upgrade procedure is to always upgrade ovn-controllers
> >> first and OVN DBs and ovn-northd later. This leads however
On 2/3/21 1:09 PM, Marcelo Ricardo Leitner wrote:
> On Wed, Feb 03, 2021 at 01:30:00PM +0800, we...@ucloud.cn wrote:
> ...
>> @@ -1641,6 +1644,10 @@ netdev_tc_flow_put(struct netdev *netdev, struct
>> match *match,
>> }
>>
>> if (mask->ct_state) {
>> +if (mask->ct_state & TC_UN
On 2/3/21 12:06 PM, Numan Siddique wrote:
On Tue, Feb 2, 2021 at 3:15 PM Dumitru Ceara wrote:
The supported upgrade procedure is to always upgrade ovn-controllers
first and OVN DBs and ovn-northd later. This leads however to the
situation when ovn-controller might try to set the Port_Binding.
On Wed, Feb 03, 2021 at 01:30:00PM +0800, we...@ucloud.cn wrote:
...
> @@ -1641,6 +1644,10 @@ netdev_tc_flow_put(struct netdev *netdev, struct match
> *match,
> }
>
> if (mask->ct_state) {
> +if (mask->ct_state & TC_UNSUPP_OVS_CS_FLAGS) {
> +return EOPNOTSUPP;
> +
On Wed, Feb 03, 2021 at 10:54:19AM +0200, Paul Blakey wrote:
>
>
> On Tue, 2 Feb 2021, Marcelo Leitner wrote:
>
> > On Tue, Feb 02, 2021 at 04:59:44PM +0100, Ilya Maximets wrote:
> > > On 2/2/21 4:52 PM, wenxu wrote:
> > > >
> > > > Hi,
> > > >
> > > > just ingore my patch. Now kernel can supp
On Tue, Feb 2, 2021 at 3:15 PM Dumitru Ceara wrote:
>
> The supported upgrade procedure is to always upgrade ovn-controllers
> first and OVN DBs and ovn-northd later. This leads however to the
> situation when ovn-controller might try to set the Port_Binding.up field
> while the Southbound DB isn
Bleep bloop. Greetings Numan Siddique, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 108 characters long (recommended limit is 79)
#165 FILE: rhel/usr_lib_system
On Sat, Jan 30, 2021 at 4:15 AM Lorenzo Bianconi
wrote:
>
> Introduce the --bfd option to lr-route-add command.
> If the BFD session UUID is provided, it will be used for the OVN route
> otherwise the next-hop will be used to perform a lookup in the OVN BFD
> table.
> If the lookup fails and outpo
On Tue, 2 Feb 2021, Marcelo Leitner wrote:
> On Tue, Feb 02, 2021 at 04:59:44PM +0100, Ilya Maximets wrote:
> > On 2/2/21 4:52 PM, wenxu wrote:
> > >
> > > Hi,
> > >
> > > just ingore my patch. Now kernel can support match invalid
> > > ct_state in th tc flower.
> >
> > I don't think that we
From: Vladislav Odintsov
New ovn-ic.service and ovn-ic-db.service systemd units were added
to manage ovn-ic and ovsdb-servers for OVN IC respectively.
ovn-ic-db.service manages both OVN_IC_Northbound and OVN_IC_Southbound
OVSDB server processes.
Submitted-at: https://github.com/ovn-org/ovn/pull/
From: Vladislav Odintsov
This reverts commit 8abe98f386ff2b4a1f2f7b61188a377a5d020429.
Submitted-at: https://github.com/ovn-org/ovn/pull/65
Signed-off-by: Vladislav Odintsov
---
rhel/automake.mk | 1 -
rhel/ovn-fedora.spec.in| 58 +++--
42 matches
Mail list logo