tc does not support conntrack ALGs. Even worse, with tc enabled, they
should not be used/configured at all. This is because even though TC
will ignore the rules with ALG configured, i.e., they will flow through
the kernel module, return traffic might flow through a tc conntrack
rule, and it will
The tc conntrack implementation does not support the timeout option.
The current implementation is silently ignoring the timeout option
by adding a general conntrack entry.
This patch will skip the related test by overriding the support macro.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
This change fixes the "conntrack - floating" test for the TC
offload case. In this scenario, the connection might move to
CLOSE_WAIT, which would fail the test as it only accepts
TIME_WAIT. However, both indicate the connection was
established, so the test should pass.
Signed-off-by: Eelco
Flush datapath flows as TC flows take some more time to be flushed out.
The flush speeds this up.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-offloads.at |2 --
tests/system-traffic.at |6 ++
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git
The goal was to run 200 successful tc tests in a row. To do this the
following was run:
for i in {1..200}; do make check-offloads || break; \
echo "ALL_200_OK: $i"; done;
Unfortunately, a bunch of test cases showed occasional failures.
For now, they are excluded from the test cases and
With some datapaths, read TC, it takes a bit longer to update the
OpenFlow statistics. Rather than adding an additional delay, try
to read the counters multiple times until we get the desired value.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-offloads.at |2 --
On 24 Jan 2023, at 7:39, Roi Dayan wrote:
> On 13/12/2022 17:35, Eelco Chaudron wrote:
>> tc does not support conntrack ALGs. Even worse, with tc enabled, they
>> should not be used/configured at all. This is because even though TC
>> will ignore the rules with ALG configured, i.e., they will
Include and run the system-traffic.at tests as part of the system offload
testsuite. Exclude all the tests that will not run without any special
modifications.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/automake.mk |1
tests/system-offloads-testsuite.at |
This series makes it possible to include system-traffic.at tests into
"make check-offloads" tests.
The last patch of the series explains which tests are still not passing
and might need some more work.
I'll try to work on the remaining failing test cases or find someone
who can work on them.
When the test description is part of the OVS_TEST_SKIP_LIST
variable, the test is skipped.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/ofproto-macros.at |5 -
tests/ovs-macros.at |7 +++
2 files changed, 11 insertions(+), 1 deletion(-)
diff --git
This patch adds a delay before dumping the conntrack table because with
tc it takes a bit longer before it gets synced.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-common-macros.at |3 +
tests/system-offloads.at | 25 +
tests/system-traffic.at | 198
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
The revalidator/purge commands in the system test cases sometimes
get called immediately after a partial test is completed. This
could cause the revalidator thread to log an error that it can
not find/delete a flow due to the slower flow installation nature
of TC.
This patch uses a macro to call
TC flower does not allow the push of the implicit null labels (RFC3032).
Avoid the use of such labels in the MPLS test cases.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-offloads.at |2 --
tests/system-traffic.at |8
2 files changed, 4 insertions(+), 6
Make the order of the Netlink attributes for odp_flow_key_from_flow__()
the same as the kernel will return them.
This will make sure the attributes displayed in the dpctl/dump-flows
output appear in the same order for all datapath.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
If a tc flow was installed but has not yet been used, report it as such.
In addition, add a delay to the "IGMP - flood under normal action" test
case to make it work with many repetitions. This delay is also present
in other ICMP/IGMP tests.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
tc does not include ethernet header length in packet byte count.
This fix will allow the packets that go trough tc to be 14 bytes less.
This difference in the TC implementation is already described in
tc-offload.rst.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
Bleep bloop. Greetings Eelco Chaudron, 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=diff' to see the
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
tc was not setting the OVS_CT_ATTR_FORCE_COMMIT flag when a forced
commit was requested. This patch will fix this.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
lib/netdev-offload-tc.c | 13 +++--
tests/system-offloads.at |1 -
2 files changed, 11 insertions(+), 3
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
Bleep bloop. Greetings Eelco Chaudron, 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.
Patch skipped due to previous failure.
Please check this out. If you feel there has been an error,
On Tue, Jan 24, 2023 at 12:28 PM Ales Musil wrote:
> To make it easier to add flows to this stage, refactor the function.
> This also has the benefit that we should see fewer allocations due to
> rearranging how we create flows and how we manipulate the match string.
>
> Signed-off-by: Ales
When the test description is part of the OVS_TEST_SKIP_LIST
variable, the test is skipped.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/ofproto-macros.at |5 -
tests/ovs-macros.at |7 +++
2 files changed, 11 insertions(+), 1 deletion(-)
diff --git
The goal was to run 200 successful tc tests in a row. To do this the
following was run:
for i in {1..200}; do make check-offloads || break; \
echo "ALL_200_OK: $i"; done;
Unfortunately, a bunch of test cases showed occasional failures.
For now, they are excluded from the test cases and
With some datapaths, read TC, it takes a bit longer to update the
OpenFlow statistics. Rather than adding an additional delay, try
to read the counters multiple times until we get the desired value.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-offloads.at |2 --
On 1/24/23 14:12, Vladislav Odintsov wrote:
> Hi Ilya,
>
> could you please take a look on this?
> Maybe you can advice any direction how to investigate this issue?
>
> Thanks in advance.
>
> Regards,
> Vladislav Odintsov
>
>> On 24 Nov 2022, at 21:10, Anton Vazhnetsov wrote:
>>
>> Hi, Terry!
Thanks Ilya for the quick and useful response!
We’ll dig into monitor/db_change_aware logic.
Regards,
Vladislav Odintsov
> On 24 Jan 2023, at 17:00, Ilya Maximets wrote:
>
> On 1/24/23 14:12, Vladislav Odintsov wrote:
>> Hi Ilya,
>>
>> could you please take a look on this?
>> Maybe you can
pmd-perf-show with pmd-perf-metrics=true displays a histogram
with averages. However, averages were not displayed when there
is no iterations.
They will be all zero so it is not hiding useful information
but the stats look incomplete without them, especially when
they are displayed for some PMD
To make it easier to add flows to this stage, refactor the function.
This also has the benefit that we should see fewer allocations due to
rearranging how we create flows and how we manipulate the match string.
Signed-off-by: Ales Musil
---
v5: Rebase on top of current main.
Add missing ");"
In order to reuse parsed data keep hash map
of 'struct ovn_controller_lb', that is local for the
controller, in separate engine node called 'lb_data'.
Those data will be later on used to determine if we need
to flush CT for changfed/removed LB backends.
Reported-at:
Remove CT for LB backends that are affected by any
LB change. This way there shouldn't be any stale entry
in the CT. This feature is enabled by default but is dependent
on OvS support for CT flush via OFP extension.
Reported-at: https://bugzilla.redhat.com/1839103
Signed-off-by: Ales Musil
---
From: Chava Leviatan
Tx queue are set each time port is added/removed or the cmask changes by
reconfigure_datapath. The amount of TX queues is set according to PMD
thread and does not take into consideration the device capabilities .
As a result , when transmitting packet from OVS to device
Some stats in pmd-perf-show don't check for divide by zero
which results in not a number (-nan).
This is a normal case for some of the stats when there are
no Rx queues assigned to the PMD thread core.
It is not obvious what -nan is to a user so add a check for
divide by zero and set stat to 0
Include and run the system-traffic.at tests as part of the system offload
testsuite. Exclude all the tests that will not run without any special
modifications.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/automake.mk |1
tests/system-offloads-testsuite.at |
If a tc flow was installed but has not yet been used, report it as such.
In addition, add a delay to the "IGMP - flood under normal action" test
case to make it work with many repetitions. This delay is also present
in other ICMP/IGMP tests.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
tc does not support conntrack ALGs. Even worse, with tc enabled, they
should not be used/configured at all. This is because even though TC
will ignore the rules with ALG configured, i.e., they will flow through
the kernel module, return traffic might flow through a tc conntrack
rule, and it will
The tc conntrack implementation does not support the timeout option.
The current implementation is silently ignoring the timeout option
by adding a general conntrack entry.
This patch will skip the related test by overriding the support macro.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
Hi Ilya,
could you please take a look on this?
Maybe you can advice any direction how to investigate this issue?
Thanks in advance.
Regards,
Vladislav Odintsov
> On 24 Nov 2022, at 21:10, Anton Vazhnetsov wrote:
>
> Hi, Terry!
>
> In continuation to our yesterday’s conversation [0], we were
On 1/24/23 12:40, clevia...@marvell.com wrote:
> From: Chava Leviatan
>
> Tx queue are set each time port is added/removed or the cmask changes by
> reconfigure_datapath. The amount of TX queues is set according to PMD
> thread and does not take into consideration the device capabilities .
> As
Make the order of the Netlink attributes for odp_flow_key_from_flow__()
the same as the kernel will return them.
This will make sure the attributes displayed in the dpctl/dump-flows
output appear in the same order for all datapath.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tc does not include ethernet header length in packet byte count.
This fix will allow the packets that go trough tc to be 14 bytes less.
This difference in the TC implementation is already described in
tc-offload.rst.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
This patch adds a delay before dumping the conntrack table because with
tc it takes a bit longer before it gets synced.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-common-macros.at |3 +
tests/system-offloads.at | 25 +
tests/system-traffic.at | 199
The revalidator/purge commands in the system test cases sometimes
get called immediately after a partial test is completed. This
could cause the revalidator thread to log an error that it can
not find/delete a flow due to the slower flow installation nature
of TC.
This patch uses a macro to call
tc was not setting the OVS_CT_ATTR_FORCE_COMMIT flag when a forced
commit was requested. This patch will fix this.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
lib/netdev-offload-tc.c | 13 +++--
tests/system-offloads.at |1 -
2 files changed, 11 insertions(+), 3
This change fixes the "conntrack - floating" test for the TC
offload case. In this scenario, the connection might move to
CLOSE_WAIT, which would fail the test as it only accepts
TIME_WAIT. However, both indicate the connection was
established, so the test should pass.
Signed-off-by: Eelco
Flush datapath flows as TC flows take some more time to be flushed out.
The flush speeds this up.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-offloads.at |2 --
tests/system-traffic.at |6 ++
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git
On 1/20/23 10:49, Simon Horman wrote:
> Hi Han Ding,
>
> On Thu, Jan 05, 2023 at 02:15:03PM +0800, Han Ding wrote:
>> Function is_gratuitous_arp() and function is_garp() are all used to judge
>> whether the flow is gratuitous arp. It is not necessary to use two functions
>> to do the same thing
This series makes it possible to include system-traffic.at tests into
"make check-offloads" tests.
The last patch of the series explains which tests are still not passing
and might need some more work.
I'll try to work on the remaining failing test cases or find someone
who can work on them.
On 24 Jan 2023, at 10:07, Eelco Chaudron wrote:
> This series makes it possible to include system-traffic.at tests into
> "make check-offloads" tests.
>
> The last patch of the series explains which tests are still not passing
> and might need some more work.
>
> I'll try to work on the
TC flower does not allow the push of the implicit null labels (RFC3032).
Avoid the use of such labels in the MPLS test cases.
Signed-off-by: Eelco Chaudron
Acked-by: Roi Dayan
---
tests/system-offloads.at |2 --
tests/system-traffic.at |8
2 files changed, 4 insertions(+), 6
Thanks for the review, Dumitru. I added my sign-off and pushed the
change to main.
On 1/19/23 04:41, Dumitru Ceara wrote:
On 1/16/23 14:33, Mark Michelson wrote:
---
Documentation/internals/release-process.rst | 20 ++--
1 file changed, 10 insertions(+), 10 deletions(-)
From: Adrian Moreno
Adding a OFPACT_SAMPLE action to a flow is useless (and even detrimental
in terms of performance) if a Flow_Sample_Collector_Set row does not
exist with the correspondent id. The sample (i.e: upcall) would take
place but ovs-vswitchd would not have a target IPFIX collector to
From: Adrian Moreno
The engine handler needs to return "false" in order to trigger a full
recompute of the physical flows.
Fixes: a42c808f30b4 ("northd: add drop sampling")
Signed-off-by: Adrian Moreno
---
controller/ovn-controller.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
From: Adrian Moreno
Some drops were left unsampled and a change in IFS made the test fail to
detect them. This patch fixes it.
Fixes: a42c808f30b4 ("northd: add drop sampling")
Acked-by: Mark Michelson
Signed-off-by: Adrian Moreno
---
northd/northd.c | 17 ++---
tests/ovn.at|
From: Adrian Moreno
The default zero value would likely match an existing openflow port and
end up generating a sample with wrong output interface information.
Since in this case we're sampling in the middle of the pipeline, the
correct value for sampling port is OFP_NONE.
Fixes: a42c808f30b4
While testing, I discovered some problems with drop sampling (first 4
patches).
Also, this series introduces an optimization. In order to avoid adding
sample actions on Chassis that do not have a Flow_Sample_Collector_Set
configured (which would generate a useless upcall), make the controller
From: Adrian Moreno
Make physical (pflow) engine node also depend on
Flow_Sample_Collector_Set table and only enable flow sampling if the
right collector set exists.
Signed-off-by: Adrian Moreno
---
controller/ovn-controller.c | 104 ++--
1 file changed, 75
From: Adrian Moreno
The drop was implicit (using empty actions). Make it explicit and
sampled.
Fixes: a42c808f30b4 ("northd: add drop sampling")
Acked-by: Mark Michelson
Signed-off-by: Adrian Moreno
---
controller/physical.c | 1 +
1 file changed, 1 insertion(+)
diff --git
On 08/01/2023 03:55, Cheng Li wrote:
In my test, if one logical core is pinned to PMD thread while the
other logical(of the same physical core) is not. The PMD
performance is affected the by the not-pinned logical core load.
This maks it difficult to estimate the loads during a dry-run.
On 1/23/23 16:36, Lorenzo Bianconi wrote:
>> On Mon, 2023-01-23 at 10:01 +0100, Lorenzo Bianconi wrote:
>>> Avoid learning Link-Local reserved multicast addresses if advertised in
>>> a MLD reports since this interferes with Slaac IPv6 address resolution
>>> implemented in OVN.
>>>
>>>
On Mon, Jan 23, 2023 at 3:50 PM Simon Horman wrote:
>
> On Mon, Jan 23, 2023 at 03:29:50PM +0100, David Marchand wrote:
> > On Mon, Jan 23, 2023 at 3:05 PM Ilya Maximets wrote:
> > >
> > > Test fails is 'nc' is not available, it should be skipped instead.
> > >
> >
> > Probably not important,
From: Adrian Moreno
IPFIX templates have to be sent for each Observation Domain ID.
Currently, a timer is kept at each dpif_ipfix_exporter to send them.
This works fine for per-bridge sampling where there is only one
Observation Domain ID per exporter. However, this is does not work for
per-flow
From: Adrian Moreno
Add options to the IPFIX table configure the interval to send statistics
and template information.
Signed-off-by: Adrian Moreno
---
- v2:
- Fixed a potential race condition in unit test.
- v1:
- Added unit test.
---
NEWS | 2 ++
Bleep bloop. Greetings Adrián Moreno, 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: Comment with 'xxx' marker
#271 FILE: ofproto/ofproto-dpif-ipfix.c:2925:
/*
> On 1/23/23 16:36, Lorenzo Bianconi wrote:
> >> On Mon, 2023-01-23 at 10:01 +0100, Lorenzo Bianconi wrote:
> >>> Avoid learning Link-Local reserved multicast addresses if advertised in
> >>> a MLD reports since this interferes with Slaac IPv6 address resolution
> >>> implemented in OVN.
> >>>
>
On Tue, Jan 24, 2023 at 2:21 PM Adrián Moreno wrote:
>
> From: Adrian Moreno
>
> IPFIX templates have to be sent for each Observation Domain ID.
> Currently, a timer is kept at each dpif_ipfix_exporter to send them.
> This works fine for per-bridge sampling where there is only one
> Observation
72 matches
Mail list logo
