Thanks!
Regards,
Vladislav Odintsov
> On 16 Jun 2021, at 02:52, Numan Siddique wrote:
>
> On Tue, Jun 15, 2021 at 2:41 PM Vladislav Odintsov wrote:
>>
>> I’m sorry for the broken tests, for some reason I sent wrong revision of
>> them.
>> Thanks you for fixing them.
>>
>> One of the
When vlan-passthru is on, VIFs may attach different VLAN tags. In this
case, VIFs are not guaranteed to belong to the same L2 broadcast domain.
Because of that, we don't know if a peer port on the switch has the same
tag used and should not allow the local responder to generate neighbour
traffic.
Bleep bloop. Greetings Gaetan Rivet, 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 97 characters long (recommended limit is 79)
#190 FILE: lib/mpsc-queue.c:107:
On Tue, Jun 15, 2021 at 2:41 PM Vladislav Odintsov wrote:
>
> I’m sorry for the broken tests, for some reason I sent wrong revision of them.
> Thanks you for fixing them.
>
> One of the patches is a bugfix, another one is a small optimisation. I know
> you
> backport only fixes without new
On Tue, Jun 15, 2021 at 3:02 PM William Tu wrote:
>
> On Tue, Jun 15, 2021 at 11:43 AM Ansis wrote:
> >
> > On Mon, Jun 14, 2021 at 10:22 PM Ansis wrote:
> > >
> > > On Mon, Jun 7, 2021 at 1:31 PM Toms Atteka wrote:
> > > >
> > > > This optimization caused FLOW_TNL_F_UDPIF flag not to be used
A lock is taken during conn_lookup() to check whether a connection is
expired before returning it. This lock can have some contention.
Even though this lock ensures a consistent sequence of writes, it does
not imply a specific order. A ct_clean thread taking the lock first
could read a value that
The lock priority order is for the global 'ct_lock' to be taken first
and then 'conn->lock'. This is an issue, as multiple operations on
connections are thus blocked between threads contending on the
global 'ct_lock'.
This was previously necessary due to how the expiration lists, timeout
policies
Multiple lookups are done to stored timeout policies, each time blocking
the global 'ct_lock'. This is usually not necessary and it should be
acceptable to get policy updates slightly delayed (by one RCU sync
at most). Using a CMAP reduces multiple lock taking and releasing in
the connection
The atomic exchange operation is a useful primitive that should be
available as well. Most compiler already expose or offer a way
to use it, but a single symbol needs to be defined.
Signed-off-by: Gaetan Rivet
Reviewed-by: Eli Britstein
---
lib/ovs-atomic-c++.h | 3 +++
Change the data structure from hmap to cmap for zone limits.
As they are shared amongst multiple conntrack users, multiple
readers want to check the current zone limit state before progressing in
their processing. Using a CMAP allows doing lookups without taking the
global 'ct_lock', thus reducing
Add a lockless multi-producer/single-consumer (MPSC), linked-list based,
intrusive, unbounded queue that does not require deferred memory
management.
The queue is designed to improve the specific MPSC setup. A benchmark
accompanies the unit tests to measure the difference in this configuration.
Change the connection expiration lists from ovs_list to mpsc-queue.
This is a pre-step towards reducing the granularity of 'ct_lock'.
It simplifies the responsibilities toward updating the expiration queue.
The dataplane now appends the new conn for expiration once during
creation. Any further
Conntracks are executed within the datapath. Locks along this path are crucial
and their critical section should be minimal. The global 'ct_lock' is necessary
before any action taken on connection states. This lock is needed for many
operations on the conntrack, slowing down the datapath.
The
The 'hash_basis' field is used sometimes during sub-systems init
routine. It will be 0 by default before randomization. Sub-systems would
then init some nodes with incorrect hash values.
The timeout policies module is affected, making the default policy being
referenced using an incorrect hash
Addresses the following BZ:
https://bugzilla.redhat.com/show_bug.cgi?id=1964953
Fabrizio D'Angelo (2):
northd: Port dummy-numa cli option from OVS
ovn-macros.at: Enable northd parallelization
northd/ovn-northd-ddlog.c | 9 -
northd/ovn-northd.c | 7 +++
- Add NORTHD_USE_PARALLELIZATION variable to enable it in ovn_start.
- Add NORTHD_DUMMY_NUMA variable for passing --dummy-numa flag when
tests are run in a low CPU system.
Signed-off-by: Fabrizio D'Angelo
---
tests/ovn-macros.at | 40 +++-
1 file changed,
Signed-off-by: Fabrizio D'Angelo
---
northd/ovn-northd-ddlog.c | 9 -
northd/ovn-northd.c | 7 +++
2 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/northd/ovn-northd-ddlog.c b/northd/ovn-northd-ddlog.c
index 444e13e39..d81414ce7 100644
---
Currently there is an option to add/flush/show ARP/ND neighbor. This covers L3
side. For L2 side, there is only fdb show command. This patch gives an option
to add/del an fdb entry via ovs-appctl.
CLI command looks like:
To add:
ovs-appctl fdb/add
ovs-appctl fdb/add br0 p1 0
On Fri, Jun 11, 2021 at 6:11 AM Dumitru Ceara wrote:
>
> This depends on the following OVS patch to be accepted:
>
> https://patchwork.ozlabs.org/project/openvswitch/patch/162264080113.315078.1220132318734842720.stgit@ebuild/
>
> However, the capability name, "ct_zero_snat" has been agreed upon
On 6/11/21 4:29 PM, Terry Wilson wrote:
> This is a partial revert of c1aa16d19, but keeps its test.
>
> Applications that use eventlet cannot use poll() without blocking.
> To fix an issue where the check_connection_completion() code would
> not detect connection errors when using select(), we
On Tue, Jun 15, 2021 at 11:43 AM Ansis wrote:
>
> On Mon, Jun 14, 2021 at 10:22 PM Ansis wrote:
> >
> > On Mon, Jun 7, 2021 at 1:31 PM Toms Atteka wrote:
> > >
> > > This optimization caused FLOW_TNL_F_UDPIF flag not to be used in
> > > hash calculation for geneve tunnel when revalidating flows
On Mon, Jun 14, 2021 at 10:22 PM Ansis wrote:
>
> On Mon, Jun 7, 2021 at 1:31 PM Toms Atteka wrote:
> >
> > This optimization caused FLOW_TNL_F_UDPIF flag not to be used in
> > hash calculation for geneve tunnel when revalidating flows which
> > resulted in different cache hash values and
I’m sorry for the broken tests, for some reason I sent wrong revision of them.
Thanks you for fixing them.
One of the patches is a bugfix, another one is a small optimisation. I know you
backport only fixes without new behaviour. I’m okay with this, but if you can
backport both, it would be less
On Tue, Jun 15, 2021 at 12:27 PM Vladislav Odintsov wrote:
>
> This series fixes duplicated MMR records in VTEP Database, which
> previously was created for each port binding.
> Also, now VTEP Logical_Switch's tunnel_key and replication_mode
> are committed only if needed.
Thanks Vladislav for
On Thu, Jun 10, 2021 at 8:48 AM Dumitru Ceara wrote:
>
> There's no need to parse the IP sets every time we iterate through them.
> It's enough to parse them once for every main loop iteration.
>
> Reported-at: https://bugzilla.redhat.com/1962338
> Signed-off-by: Dumitru Ceara
Thanks for the
On Tue, Jun 15, 2021 at 9:28 AM Mark Michelson wrote:
>
> On 6/15/21 9:24 AM, Mark Michelson wrote:
> > Signed-off-by: Mark Michelson
For both the patches in the series:
Acked-by: Numan Siddique
> > ---
> > controller/ofctrl.c | 1 -
> > 1 file changed, 1 deletion(-)
> >
> > diff --git
> Hi Ian,
>
> Thanks for the review. My responses are inline.
>
> > -Original Message-
> > From: Stokes, Ian
> > Sent: Tuesday 1 June 2021 19:58
> > To: Ferriter, Cian ; ovs-dev@openvswitch.org; Van
> Haaren, Harry
> > Cc: i.maxim...@ovn.org
> > Subject: RE: [ovs-dev] [v12 03/16]
> Hi Ian,
>
> Thanks for the review. My responses are inline.
>
> > -Original Message-
> > From: Stokes, Ian
> > Sent: Tuesday 1 June 2021 19:58
> > To: Ferriter, Cian ; ovs-dev@openvswitch.org; Van
> Haaren, Harry
> > Cc: i.maxim...@ovn.org; Gaetan Rivet ;
> el...@nvidia.com
> >
> Hi Ian,
>
> Thanks for the review. My responses are inline.
>
> > -Original Message-
> > From: Stokes, Ian
> > Sent: Wednesday 19 May 2021 16:54
> > To: Ferriter, Cian ; ovs-dev@openvswitch.org
> > Cc: i.maxim...@ovn.org
> > Subject: RE: [ovs-dev] [v12 01/16] dpif-netdev: Refactor to
> From: Harry van Haaren
>
> This commit optimizes the output action, by enabling the compiler to
> optimize the code better through reducing code complexity.
>
> The core concept of this optimization is that the array-length checks
> have already been performed above the copying code, so can
> From: Harry van Haaren
>
> Optimize for the best case here where all packets will be compatible
> with 'netdev_flags'.
>
> Signed-off-by: Harry van Haaren
> Co-authored-by: Cian Ferriter
> Signed-off-by: Cian Ferriter
Thanks for the patch Cian/Harry, comments inline.
In general I have a
Hi Dumitru,
Thanks for the review, new patch is submitted to the list.
Regards,
Vladislav Odintsov
> On 15 Jun 2021, at 18:50, Dumitru Ceara wrote:
>
> On 6/11/21 1:44 PM, Vladislav Odintsov wrote:
>> Before this patch ovn-controller-vtep created Mcast_Macs_Remote
>> record for each Port
Signed-off-by: Vladislav Odintsov
Acked-by: Dumitru Ceara
---
controller-vtep/vtep.c | 19 ---
1 file changed, 12 insertions(+), 7 deletions(-)
diff --git a/controller-vtep/vtep.c b/controller-vtep/vtep.c
index 49723b39d..314cc4eae 100644
--- a/controller-vtep/vtep.c
+++
Before this patch ovn-controller-vtep created Mcast_Macs_Remote
record for each Port Binding in the ovn Logical Switch, to which
vtep Logical Switch was attached.
With this patch there is only one Mcast_Macs_Remote record per datapath.
Physical Locator set is created every time when physical
This series fixes duplicated MMR records in VTEP Database, which
previously was created for each port binding.
Also, now VTEP Logical_Switch's tunnel_key and replication_mode
are committed only if needed.
v3 -> v4
* Deleted form feedes returned back.
v2 -> v3
* Addressed review
On Tue, Jun 1, 2021 at 9:33 AM Dumitru Ceara wrote:
>
> There's no need to allocate a logical flow structure if we're going to
> merge it to an existing one that refers to a datapath group.
>
> This saves a lot of xstrdup() calls followed immediately by free().
>
> Reported-at:
On 6/15/21 3:43 PM, Alin Gabriel Serdean wrote:
> VStudio 16.10 adds usermode includes before including the driver kit ones.
>
> Bug tracked at:
> https://developercommunity.visualstudio.com/t/error-lnk2019-unresolved-external-symbol-stdio-com/1434674
>
> Fixes appveyor build reported by forcing
Fixes: 1ca0323e7c29 ("Require Python 3 and remove support for Python 2.")
Reported at: https://bugzilla.redhat.com/show_bug.cgi?id=1949875
Signed-off-by: Rosemarie O'Riorden
---
Version 2:
- Fixed import errors in v1.
- Removed lines that indicate compatibility with python 2.
Version 3:
-
On 6/11/21 1:44 PM, Vladislav Odintsov wrote:
> Signed-off-by: Vladislav Odintsov
> ---
Acked-by: Dumitru Ceara
Thanks!
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On 6/11/21 1:44 PM, Vladislav Odintsov wrote:
> Before this patch ovn-controller-vtep created Mcast_Macs_Remote
> record for each Port Binding of the OVN Logical Switch, to which
> VTEP Logical Switch was attached.
> With this patch there is only one Mcast_Macs_Remote record per datapath.
>
DHCP option Hostname (opcode 12) can be configured in
Logical_Switch_Port as well as in DHCP_Options table's
options:hostname column. If both methods of configuration were
used, Logical_Switch_Port's setting takes precedence.
Signed-off-by: Vladislav Odintsov
---
v2 -> v3
* Fixed style warning.
Bleep bloop. Greetings Vladislav Odintsov, 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 has trailing whitespace
#87 FILE: ovn-nb.xml:3320:
On Tue, Jun 15, 2021 at 10:30 AM Vladislav Odintsov wrote:
>
> Glad to hear that.
>
> I’ve sent a new patch version:
> https://patchwork.ozlabs.org/project/ovn/patch/20210615142405.20140-1-odiv...@gmail.com/
> BTW, is there any other way to update patch except posting a complete new
> patch to
Glad to hear that.
I’ve sent a new patch version:
https://patchwork.ozlabs.org/project/ovn/patch/20210615142405.20140-1-odiv...@gmail.com/
BTW, is there any other way to update patch except posting a complete new patch
to patchwork with vN version prefix? Sorry if it’s a stupid question, I’m
DHCP option Hostname (opcode 12) can be configured in
Logical_Switch_Port as well as in DHCP_Options table's
options:hostname column. If both methods of configuration were
used, Logical_Switch_Port's setting takes precedence.
Signed-off-by: Vladislav Odintsov
---
v1 -> v2
* Added support for
VStudio 16.10 adds usermode includes before including the driver kit ones.
Bug tracked at:
https://developercommunity.visualstudio.com/t/error-lnk2019-unresolved-external-symbol-stdio-com/1434674
Fixes appveyor build reported by forcing external includes.
Reported-by: Ilya Maximets
On 15/06/2021 14:33, Numan Siddique wrote:
On Tue, Jun 15, 2021 at 8:34 AM Brendan Doyle wrote:
On 15/06/2021 13:26, Numan Siddique wrote:
On Tue, Jun 15, 2021 at 7:50 AM Brendan Doyle wrote:
On 14/06/2021 21:17, Numan Siddique wrote:
On Mon, Jun 14, 2021 at 5:19 AM Brendan Doyle
On 14 Jun 2021, at 20:15, Ilya Maximets wrote:
> On 6/14/21 4:28 PM, Eelco Chaudron wrote:
>> For patch ports, the is_last_action value is not propagated and is
>> always set to true. This causes non-reversible actions to modify the
>> packet, and the original content is not preserved when
On Tue, Jun 15, 2021 at 8:34 AM Brendan Doyle wrote:
>
>
>
> On 15/06/2021 13:26, Numan Siddique wrote:
> > On Tue, Jun 15, 2021 at 7:50 AM Brendan Doyle
> > wrote:
> >>
> >>
> >> On 14/06/2021 21:17, Numan Siddique wrote:
> >>> On Mon, Jun 14, 2021 at 5:19 AM Brendan Doyle
> >>> wrote:
>
On 6/15/21 9:24 AM, Mark Michelson wrote:
Signed-off-by: Mark Michelson
---
controller/ofctrl.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/controller/ofctrl.c b/controller/ofctrl.c
index 053631590..48d001506 100644
--- a/controller/ofctrl.c
+++ b/controller/ofctrl.c
@@ -1259,7
These strings are initialized and destroyed, but they aren't used
otherwise. This commit simply removes them.
Signed-off-by: Mark Michelson
---
northd/ovn-northd.c | 5 -
1 file changed, 5 deletions(-)
diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c
index d872f6a3c..a6cea285b 100644
Signed-off-by: Mark Michelson
---
controller/ofctrl.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/controller/ofctrl.c b/controller/ofctrl.c
index 053631590..48d001506 100644
--- a/controller/ofctrl.c
+++ b/controller/ofctrl.c
@@ -1259,7 +1259,6 @@ ofctrl_flood_remove_flows(struct
On 5/21/21 7:59 PM, Aaron Conole wrote:
> As reported by Wang Liang, the way packets are passed to the ipf module
> doesn't allow for use later on in reassembly. Such packets may be get
> released anyway, such as during cleanup of tx processing. Because the
> ipf module lacks a way of forcing
Due to a bug in glibc [0], M_TRIM_THRESHOLD (default 128KB) is not
honored. The lflow cache is one of the largest memory consumers in
ovn-controller and it used to trim memory whenever the cache was
flushed. However, that required periodic intervention from the CMS
side.
Instead, we now
Signed-off-by: Dumitru Ceara
---
controller/chassis.c | 87 --
1 file changed, 28 insertions(+), 59 deletions(-)
diff --git a/controller/chassis.c b/controller/chassis.c
index 9b0a36cf0..80d516f49 100644
--- a/controller/chassis.c
+++
This is the v2 revision of the patch originally posted at:
http://patchwork.ozlabs.org/project/ovn/patch/20210604100047.24979-1-dce...@redhat.com/
Changes in v2:
- Addressed Mark Gray's comments:
- Made trim flow limit configurable.
- Made trim high watermark percentage configurable.
-
On 6/11/21 4:48 PM, Mark Gray wrote:
> On 11/06/2021 10:59, Dumitru Ceara wrote:
>> Thanks for the review!
>>
>> On 6/11/21 11:22 AM, Mark Gray wrote:
>>> On 04/06/2021 11:00, Dumitru Ceara wrote:
Due to a bug in glibc [0], M_TRIM_THRESHOLD (default 128KB) is not
honored. The lflow
On Tue, Jun 15, 2021 at 5:00 AM Vladislav Odintsov wrote:
>
> Hi Numan,
>
> Thanks for review.
>
> I’ve been thinking about such approach to set hostnames via DHCP_Options. In
> my opinion it can be an overkill to have a special DHCP_Options record per
> each logical switch port in case where
On 15/06/2021 13:26, Numan Siddique wrote:
On Tue, Jun 15, 2021 at 7:50 AM Brendan Doyle wrote:
On 14/06/2021 21:17, Numan Siddique wrote:
On Mon, Jun 14, 2021 at 5:19 AM Brendan Doyle wrote:
Hi,
Just wondering how to move this along. It's been in the queue a while now.
The product I'm
On Tue, Jun 15, 2021 at 7:50 AM Brendan Doyle wrote:
>
>
>
> On 14/06/2021 21:17, Numan Siddique wrote:
> > On Mon, Jun 14, 2021 at 5:19 AM Brendan Doyle
> > wrote:
> >> Hi,
> >>
> >> Just wondering how to move this along. It's been in the queue a while now.
> >> The product I'm working on
Bleep bloop. Greetings Mark Gray, 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)
#44 FILE:
From: Alexey Roytman
The original version said that "monitor_cond_change" request response
should contain '"result": null'. However, if response message has form
like {"id":13, "result":null} or {"id":13} the ovsdb client
(ovn-controller) returns the folling misleading message and closes
Add a suite of micro-benchmarks to aid a developer in understanding the
performance impact of any changes that they are making. They can be used to
help to understand the relative performance between two test runs on the same
test machine, but are not intended to give the absolute performance of
For performance measurement, it is useful to understand the
length of time required to complete a number of key code paths
in ovn-northd.c. Add stopwatches to measure these timings.
Signed-off-by: Mark Gray
---
northd/ovn-northd-ddlog.c | 15 +++
northd/ovn-northd.c | 20
Signed-off-by: Mark Gray
---
tests/ovn-macros.at | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/ovn-macros.at b/tests/ovn-macros.at
index cd02b6986cc2..c92c07720112 100644
--- a/tests/ovn-macros.at
+++ b/tests/ovn-macros.at
@@ -184,7 +184,7 @@ ovn_start_northd() {
#
This is a proposal to add some micro-benchmarks to aid developers
in benchmarking optimizations to OVN. It starts by adding simple
metrics for northd but could be expanded in future patches.
Mark Gray (3):
ovn-macros.at: fix typo
ovn-northd: Add useful stopwatches
tests: Add check-perf
On 14/06/2021 21:17, Numan Siddique wrote:
On Mon, Jun 14, 2021 at 5:19 AM Brendan Doyle wrote:
Hi,
Just wondering how to move this along. It's been in the queue a while now.
The product I'm working on requires this feature and If I can't get it
upstream
I'll need to look at other options.
From: Meher Chinwala
system-dpdk: Negotiation Tests for TSO
This patch adds negotiation tests for checking whether TSO is enabled or not in
OVS and in TestPMD for 4 diferent scenarios.
Signed-off-by: Meher Chinwala
---
tests/system-dpdk-macros.at | 43 ++
On 6/14/21 11:28 PM, Rosemarie O'Riorden wrote:
> Fixes: 1ca0323e7c29 ("Require Python 3 and remove support for Python 2.")
> Reported at: https://bugzilla.redhat.com/show_bug.cgi?id=1949875
> Signed-off-by: Rosemarie O'Riorden
> ---
> Version 2:
> - Fixed import errors in v1.
> - Removed
Hi,
I’ve posted a new version of this patch:
https://patchwork.ozlabs.org/project/ovn/list/?series=248346
Regards,
Vladislav Odintsov
> On 11 Jun 2021, at 01:45, Ben Pfaff wrote:
>
> On Fri, Jun 11, 2021 at 12:13:46AM +0200, Dumitru Ceara wrote:
>> On 6/10/21 11:47 PM, Ben Pfaff wrote:
>>>
Hi Numan,
Thanks for review.
I’ve been thinking about such approach to set hostnames via DHCP_Options. In my
opinion it can be an overkill to have a special DHCP_Options record per each
logical switch port in case where we want to return hostnames in DHCP Reply. I
understand hostname as a
On 6/15/21 4:31 AM, Tony van der Peet wrote:
> I did the patch as suggested, and it looks something like this:
>
> --- a/lib/dpif-netdev.c
> +++ b/lib/dpif-netdev.c
> @@ -4168,9 +4168,13 @@dpif_netdev_execute(struct dpif *dpif, struct
> dpif_execute *execute)
>
On 14 Jun 2021, at 20:15, Ilya Maximets wrote:
> On 6/14/21 4:28 PM, Eelco Chaudron wrote:
>> For patch ports, the is_last_action value is not propagated and is
>> always set to true. This causes non-reversible actions to modify the
>> packet, and the original content is not preserved when
Bleep bloop. Greetings root, 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:
ERROR: Author meher chinwala needs to sign off.
WARNING: Unexpected sign-offs from developers who are
From: meher chinwala
This patch adds negotiation tests for checking whether TSO is enabled or not in
OVS and in TestPMD for 4 diferent scenarios.
Signed-off-by: Meher Chinwala
---
tests/system-dpdk-macros.at | 43 ++
tests/system-dpdk.at| 71
75 matches
Mail list logo
