On Wed, Oct 27, 2021 at 10:23 AM Mark Gray wrote:
>
> Add the 'inc-proc-eng' framework to northd. This does *not*
> add any incremental processing at this stage but provides the
> framework to do so. Even in this base configuration, we see an
> advantage as northd no longer processes the
On Wed, Oct 27, 2021 at 10:24 AM Mark Gray wrote:
>
> In order to remove the IDL loop variable from the engine context,
> we do not calculate the database sequence numbers incrementally.
>
> Signed-off-by: Mark Gray
> ---
> lib/inc-proc-eng.h | 2 -
> northd/en-northd.c | 3 +-
>
On Wed, Oct 27, 2021 at 10:24 AM Mark Gray wrote:
>
> Add an additional node that initially does nothing. This serves as a
> template for how to add a new node. This node is inserted after
> the northd_node.
>
> This node will be updated in a later commit to generate logical
> flows for the SBDB.
On Wed, Oct 27, 2021 at 10:24 AM Mark Gray wrote:
>
> 'struct northd_data' is used to hold the global state data for the
> incremental processing node 'en_northd'. This structure will also hold
> 'struct northd_input' which will hold references to output data from
> its input nodes. In
On 11/2/2021 9:50 AM, Vladislav Odintsov wrote:
Hi,
it’s just a quick reminder, that if you know how to debug/fix this, please let
me know.
Thanks.
Regards,
Vladislav Odintsov
Hi Vladislav,
I'm very short on time at my day job right now and haven't had a chance
to consider this issue.
On Fri, Oct 29, 2021 at 5:28 PM Mark Michelson wrote:
>
> When an ARP request reaches a logical switch that is connected to
> gateway routers, if the IP address in the ARP is an "unreachable"
> address, then the ARP is flooded out all logical switch ports.
>
> This can become an issue when a
On Tue, Nov 2, 2021 at 3:35 PM Dumitru Ceara wrote:
>
> On 11/2/21 8:30 PM, Dumitru Ceara wrote:
> > This series backports the load balancer performance improvements to
> > stable branch-21.09.
> >
> > - patch 1/3 changes the way ARP responder flows are generated for load
> > balancer VIPs,
Hi Lorenzo,
> what will it happen if pinctrl learns a different MAC for the given IP after
> the CMS has
configured the static binding? Will it be overwritten? If so I guess we need a
flag to avoid it.
1. Static binding will be overwritten by pinctrl.
2. Northd will again overwrite this and
On Tue, Nov 2, 2021, at 18:12, Paolo Valerio wrote:
> with the command is now possible to change the ageing time of the
> cache entries.
>
> For the existing entries the ageing time is updated only if the
> current expiration is greater than the new one. In any case, the next
> refresh will set it
On Tue, Nov 2, 2021, at 18:12, Paolo Valerio wrote:
Hi Paolo,
I think this commit needs more details.
I'm guessing the threads involved are the PMDs and main, are there others?
Coherency is implicit on x86 cores, but those parallel read/writes are
undefined behavior & possibly incorrect on
On Wed, Nov 3, 2021 at 9:49 AM Lorenzo Bianconi
wrote:
>
> Add missing icmp{4,6}_error flows for ingress traffic destinated to FIP
> associated to the interface.
>
> Signed-off-by: Lorenzo Bianconi
> ---
> northd/northd.c | 87 ++---
> tests/ovn.at
On Tue, Nov 2, 2021 at 11:46 PM Han Zhou wrote:
>
> It is important to keep conjunction IDs consistent between runs to avoid
> unnecessary OVS flows deletion and reinstallation. However, There are two
> problems of the current lflow-cache based conjunction id peristent
approach.
>
> 1) When
> Hi Lorenzo,
>
> Thanks for the review. Do we need a flag to control statically added MACs
> from CMS vs dynamically learnt MAC from pinctrl for a given IP?
Yes, I think so.
>
> Typically static configurations for a MAC Table override dynamically learnt
> MACs. In this patch even if pinctrl
When a logical flow translation uses conjunction ids, it is better to
keep the conjuntion ids consistent between iterations, so that the
generated OVS flows don't change unnecessarily and avoid reinstalling
unchanged flows to OVS. The problem was partially solved when
lflow-cache was enabled but
Conjunction id cache is useful to keep conjunction ids consistent between
iterations when match or expr cache is not available. However, the
current implementation caches the conjunction id whenever n_conjs > 0.
It assumes that when the lflow is reprocessed, the conjunction ids
starting from the
In commit fadbc04f38db it changed the lflow-cache behavior but forgot to
update the commment.
Fixes: fadbc04f38db ("ovn-controller: Fix incremental processing for logical
port references.")
Signed-off-by: Han Zhou
---
controller/lflow-cache.h | 5 +++--
1 file changed, 3 insertions(+), 2
It is important to keep conjunction IDs consistent between runs to avoid
unnecessary OVS flows deletion and reinstallation. However, There are two
problems of the current lflow-cache based conjunction id peristent approach.
1) When n_conjs changes, the cached conj_id_ofs can overlap with other
FAQ is updated to reflect the latest DPDK for OVS branch 2.13 and 2.14
Signed-off-by: Suneetha Kalahasthi
---
Documentation/faq/releases.rst | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Documentation/faq/releases.rst b/Documentation/faq/releases.rst
index
On Wed, Nov 3, 2021 at 10:34 AM Ilya Maximets wrote:
>
> On 10/29/21 19:26, num...@ovn.org wrote:
> > From: Numan Siddique
> >
> > xlate_check_pkt_larger() sets ctx->exit to 'true' at the end
> > causing the translation to stop. This results in incomplete
> > datapath rules.
> >
> > For example,
From: Numan Siddique
system-test matrix first runs unit tests and then system tests.
And if unit tests fail, system tests are not run.
This patch now runs only system tests for system-test matrix.
A new entry is added in the test matrix - 'gcc compiler with
unit tests'.
Signed-off-by: Numan
FAQ is updated to reflect the latest DPDK for OVS branch 2.13 and 2.14
Signed-off-by: Suneetha Kalahasthi
---
Documentation/faq/releases.rst | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Documentation/faq/releases.rst b/Documentation/faq/releases.rst
index
FAQ is updated to reflect the latest DPDK for OVS branch 2.13 and 2.14
Signed-off-by: Suneetha Kalahasthi
---
Documentation/faq/releases.rst | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Documentation/faq/releases.rst b/Documentation/faq/releases.rst
index
On 11/3/21 5:39 PM, Numan Siddique wrote:
> On Wed, Nov 3, 2021 at 12:09 PM Ilya Maximets wrote:
>>
>> On 11/3/21 17:01, num...@ovn.org wrote:
>>> From: Numan Siddique
>>>
>>> For the system-test matrix, running unit tests is not
>>> required and if a unit test fails, then system tests
>>> are
On Wed, 2021-11-03 at 16:53 +0100, Ilya Maximets wrote:
> On 10/27/21 17:48, Mike Pattrick wrote:
> > Currently the ovs-tcpdump utility creates a virtual tunnel to send
> > packets to. This method functions perfectly fine, however, it can
> > greatly impact performance of the monitored port.
> >
On Wed, Nov 3, 2021 at 12:09 PM Ilya Maximets wrote:
>
> On 11/3/21 17:01, num...@ovn.org wrote:
> > From: Numan Siddique
> >
> > For the system-test matrix, running unit tests is not
> > required and if a unit test fails, then system tests
> > are not run. Also test matrix covers unit tests
On 10/21/21 14:48, Kevin Traynor wrote:
> On 21/10/2021 08:04, Somnath Chatterjee wrote:
>> There was typo in function pointer check in
>> dpif_bond_add() before calling bond_add()
>>
>> Fixes: 9df65060cf4c ("userspace: Avoid dp_hash recirculation for balance-tcp
>> bond mode.")
>> Signed-off-by:
On 11/3/21 01:36, Terry Wilson wrote:
> Add support for monitor_cond_since / update3 to python-ovs to
> allow more efficient reconnections when connecting to clustered
> OVSDB servers.
>
> Signed-off-by: Terry Wilson
> ---
> python/ovs/db/idl.py | 55 +++-
On 11/3/21 17:01, num...@ovn.org wrote:
> From: Numan Siddique
>
> For the system-test matrix, running unit tests is not
> required and if a unit test fails, then system tests
> are not run. Also test matrix covers unit tests anyway.
Hmm. I don't think that is correct.
IIUC, the only
From: Numan Siddique
For the system-test matrix, running unit tests is not
required and if a unit test fails, then system tests
are not run. Also test matrix covers unit tests anyway.
Signed-off-by: Numan Siddique
---
.ci/linux-build.sh | 22 +++---
1 file changed, 11
On Tue, Nov 2, 2021, at 19:10, Kevin Traynor wrote:
> Previously, when a user enabled PMD auto load balancer with
> pmd-auto-lb="true", some conditions such as number of PMDs/RxQs
> that were required for a rebalance to take place were checked.
>
> If the configuration meant that a rebalance would
On 10/27/21 17:48, Mike Pattrick wrote:
> Currently the ovs-tcpdump utility creates a virtual tunnel to send
> packets to. This method functions perfectly fine, however, it can
> greatly impact performance of the monitored port.
>
> It has been reported to reduce packet throughput significantly.
On 10/12/21 21:49, David Marchand wrote:
> DPIF AVX512 optimisations currently rely on DPDK availability while
> they can be used without DPDK.
> Besides, checking for availability of some isa only has to be done once
> and won't change while a OVS process runs.
>
> Resolve isa availability in
On 11/1/21 15:27, Terry Wilson wrote:
> Acked-by: Terry Wilson
> Tested-by: Terry Wilson
>
> On Fri, Oct 29, 2021 at 6:12 PM Timothy Redaelli wrote:
>>
>> Currently, pyOpenSSL is half-deprecated upstream and so it's removed on
>> some distributions (for example on CentOS Stream 9,
>>
On 10/27/21 22:57, Terry Wilson wrote:
> On Mon, Oct 25, 2021 at 8:46 AM Timothy Redaelli wrote:
>>
>> In an upcoming patch, PyOpenSSL will be replaced with Python ssl module,
>> but in order to do an async connection with Python ssl module the ssl
>> socket must be created when the socket is
Bleep bloop. Greetings Mohammad Heib, 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 82 characters long (recommended limit is 79)
#159 FILE:
On 10/29/21 19:26, num...@ovn.org wrote:
> From: Numan Siddique
>
> xlate_check_pkt_larger() sets ctx->exit to 'true' at the end
> causing the translation to stop. This results in incomplete
> datapath rules.
>
> For example, for the below OF rules configured on a bridge,
>
> table=0,in_port=1
Hi Numan,
thank you for reviewing the patch.
On 11/3/21 1:57 AM, Numan Siddique wrote:
On Wed, Sep 22, 2021 at 1:58 PM wrote:
From: Mohammad Heib
currently ovn-northd only handle virtual ports with VIP IPv4 and ignores
virtual ports with VIP IPv6.
This patch adds support for virtual ports
Add missing icmp{4,6}_error flows for ingress traffic destinated to FIP
associated to the interface.
Signed-off-by: Lorenzo Bianconi
---
northd/northd.c | 87 ++---
tests/ovn.at| 58 ++---
2 files changed, 121
currently ovn-northd only handle virtual ports with VIP IPv4 and ignores
virtual ports with VIP IPv6.
This patch adds support for virtual ports with VIP IPv6 by adding
lflows to the lsp_in_arp_rsp logical switch pipeline.
Those lflows handle Neighbor Solicitations and Neighbor Advertisement
Problem Statement:
OVS flushes and subsequently repopulates its route
cache whenever it receives a netlink notification
about kernel interface change. At the same time the
port addition triggers a revalidation of all
datapath flow cache entries. The revalidation of
egress tunnel flows depends on
On 28/10/2021 12:22, Suneetha Kalahasthi wrote:
FAQ is updated to reflect the latest DPDK for OVS branch 2.13 and 2.14
Signed-off-by: Suneetha Kalahasthi
---
Documentation/faq/releases.rst | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/Documentation/faq/releases.rst
On 3 Nov 2021, at 9:34, Van Haaren, Harry wrote:
>> -Original Message-
>> From: Eelco Chaudron
>> Sent: Tuesday, November 2, 2021 9:10 AM
>> To: Amber, Kumar
>> Cc: ovs-dev@openvswitch.org; f...@sysclose.org; Van Haaren, Harry
>> ; Stokes, Ian ;
>> i.maxim...@ovn.org
>> Subject: Re:
On 10/21/2021 4:27 PM, Eelco Chaudron wrote:
On 21 Oct 2021, at 10:00, Chris Mi wrote:
On 10/15/2021 9:42 PM, Eelco Chaudron wrote:
Small comments inline, and Ilya please take a look at the first comment/request.
//Eelco
On 12 Oct 2021, at 10:19, Chris Mi wrote:
Some offload actions
> -Original Message-
> From: Eelco Chaudron
> Sent: Tuesday, November 2, 2021 9:10 AM
> To: Amber, Kumar
> Cc: ovs-dev@openvswitch.org; f...@sysclose.org; Van Haaren, Harry
> ; Stokes, Ian ;
> i.maxim...@ovn.org
> Subject: Re: [PATCH v3] dpcls: Change info-get function to fetch dpcls
Bleep bloop. Greetings Han Zhou, 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: sha1 information is lacking or useless (controller/lflow-cache.h).
error: could not build fake
Modify ci linux build script to use the latest DPDK stable release 19.11.10.
Modify Documentation to use the latest DPDK stable release 19.11.10.
Update NEWS file to reflect the latest DPDK stable release 19.11.10.
FAQ is updated to reflect the latest DPDK for each OVS branch.
Signed-off-by:
Modify ci linux build script to use the latest DPDK stable release 19.11.10.
Modify Documentation to use the latest DPDK stable release 19.11.10.
Update NEWS file to reflect the latest DPDK stable release 19.11.10.
FAQ is updated to reflect the latest DPDK for each OVS branch.
Signed-off-by:
When a logical flow translation uses conjunction ids, it is better to
keep the conjuntion ids consistent between iterations, so that the
generated OVS flows don't change unnecessarily and avoid reinstalling
unchanged flows to OVS. The problem was partially solved when
lflow-cache was enabled but
Conjunction id cache is useful to keep conjunction ids consistent between
iterations when match or expr cache is not available. However, the
current implementation caches the conjunction id whenever n_conjs > 0.
It assumes that when the lflow is reprocessed, the conjunction ids
starting from the
It is important to keep conjunction IDs consistent between runs to avoid
unnecessary OVS flows deletion and reinstallation. However, There are two
problems of the current lflow-cache based conjunction id peristent approach.
1) When n_conjs changes, the cached conj_id_ofs can overlap with other
50 matches
Mail list logo