On 18/02/2021 22:36, Mark Michelson wrote:
Hi Anton, in short:
Acked-by: Mark Michelson
I think this can go in as-is. My only question is with regards to how
dp_groups are handled. The current patch makes it so that when
dp_groups are used, we parallelize the work. However, instead of each
On Fri, Feb 19, 2021 at 12:53 AM Mark Gray wrote:
>
> On 18/02/2021 19:00, num...@ovn.org wrote:
> > From: Numan Siddique
> >
> > When a Gateway router is configured with a load balancer
> > and it is also configured with options:lb_force_snat_ip=,
> > OVN after load balancing the destination IP
Hi,
Looks like this bug is caused by violating the fact that if a rule is
referenced, the related ofproto should not be destroyed.
If so, I have a patch that also fixes the problem, not sure if this helps.
Aside from the comments on patch 1,
Acked-by: Mark Michelson
On 2/5/21 1:56 AM, num...@ovn.org wrote:
From: Numan Siddique
This series adds the support to learn the macs seen on logical port
VIF's whose port security is disabled and 'unknown' address set.
A new table - FDB is added in
Acked-by: Mark Michelson
On 2/12/21 2:21 AM, num...@ovn.org wrote:
From: Numan Siddique
If an external logical port 'sw0-p1' is associated with an
ha_chassis_group and later CMS runs a command to clear the
ha_chassis_group and set the type to normal in the same
transaction, i.e
ovn-nbctl
On Wed, Feb 17, 2021 at 1:09 PM Yifeng Sun wrote:
>
> ovs-vswitchd could crash under these circumstances:
> 1. When one bridge is being destroyed, ofproto_destroy() is called and
> connmgr pointer of its ofproto struct is nullified. This ofproto struct is
> deallocated through
This passed in the ovsrobot CI:
https://github.com/ovsrobot/ovn/actions/runs/579307688
I think that we have arrived at consensus to push this to ovn master
after branching happens tomorrow. Please, let me know if I
misunderstood!
___
dev mailing list
Hi!
I've found an old bug in FreeBSD bugzilla:
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=238331
reporting an incorrect behavior of ovs due to conflict of
uuid_from_string() defined in lib/uuid.* with the function with
same name from FreeBSD standard C library:
void
Hi Anton, in short:
Acked-by: Mark Michelson
I think this can go in as-is. My only question is with regards to how
dp_groups are handled. The current patch makes it so that when dp_groups
are used, we parallelize the work. However, instead of each thread
creating discrete segments of
Bleep bloop. Greetings Ben Pfaff, 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
#3724 FILE: northd/ovn-northd-ddlog.c:166:
.cs =
Bleep bloop. Greetings Ben Pfaff, 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: Unexpected sign-offs from developers who are not authors or co-authors
or committers:
The idea is to run each test twice, once with ovn-northd, once
with ovn-northd-ddlog. To do that, we add a macro OVN_FOR_EACH_NORTHD
and bracket each test that uses ovn-northd in it.
Signed-off-by: Ben Pfaff
Acked-by: Dumitru Ceara
---
tests/ovn-ic.at | 11 +-
tests/ovn-macros.at | 96
The code here checked whether 'bfd_e' was null even though that wasn't
necessary given the control flow just above it.
Signed-off-by: Ben Pfaff
---
northd/ovn-northd.c | 16
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/northd/ovn-northd.c b/northd/ovn-northd.c
I don't know a good reason to intentionally create an empty multicast
group, but disallowing empty multicast groups has an odd side effect:
you can delete all but one of the ports that a group contains and the
database will happily remove the ports from the group automatically
(because 'ports' is
Otherwise this test is racy because it assumes that northd finishes
its updates between the ovn-nbctl changes and the subsequent checks.
Also, simplify some series of "check_column" into single
"check_row_count" calls (that then become "wait_row_count").
Signed-off-by: Ben Pfaff
---
From: Leonid Ryzhyk
Export `ddlog_warn` and `ddlog_err` functions that are just wrappers
around `VLOG_WARN` and `VLOG_ERR`. This is not ideal because the
functions are exported by `ovn_util.c` and the resulting log messages use
`ovn_util` as module name. More importantly, these functions do
v10.1:
- Fold in fix to northd makefile.
v10:
- Fix linking problem reported by Dumitru and others.
- Update to ddlog 0.36.0 (thanks Leonid!).
- Port OVN commits up to 858d1dd716db ("ofctrl: Fix the assert seen when
flood removing flows.")
- Fix IGMP and MLD snoop/querier/reply tests
On 18/02/2021 19:00, num...@ovn.org wrote:
> From: Numan Siddique
>
> When a Gateway router is configured with a load balancer
> and it is also configured with options:lb_force_snat_ip=,
> OVN after load balancing the destination IP to one of the
> backend also does a NAT on the source ip with
From: Numan Siddique
When a Gateway router is configured with a load balancer
and it is also configured with options:lb_force_snat_ip=,
OVN after load balancing the destination IP to one of the
backend also does a NAT on the source ip with the
lb_force_snat_ip if the packet is destined to a load
v10.1:
- Fold in fix to northd makefile.
v10:
- Fix linking problem reported by Dumitru and others.
- Update to ddlog 0.36.0 (thanks Leonid!).
- Port OVN commits up to 858d1dd716db ("ofctrl: Fix the assert seen when
flood removing flows.")
- Fix IGMP and MLD snoop/querier/reply tests
This had a bug in the makefile. I fixed it:
https://github.com/blp/ovs-reviews/actions/runs/579105628
I'll post a v10.1.
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
Depending on the logical flow matches, multiple SB flows can point to
the same desired flow. If it happens that the desired flow conflicts
with a more restrictive (already installed) flow, then we shouldn't try
to add the desired flow multiple times to the list maintained for the
installed flow.
On Thu, Feb 18, 2021 at 11:56 PM Mark Gray wrote:
>
> On 18/02/2021 17:26, Numan Siddique wrote:
> > On Thu, Feb 18, 2021 at 9:15 PM Mark Gray wrote:
> >>
> >> On 18/02/2021 14:30, Numan Siddique wrote:
> >>> On Wed, Feb 17, 2021 at 7:37 PM Mark Gray wrote:
>
> Thanks Numan, some
If a Port_Binding corresponding to a VIF changes in the Southbound to
represent a port that cannot be bound locally (e.g., localport,
localnet, patch) then ovn-controller should not try to process this
update incrementally. If it would try, it would need to first remove
cleanup all previous
On 18/02/2021 17:26, Numan Siddique wrote:
> On Thu, Feb 18, 2021 at 9:15 PM Mark Gray wrote:
>>
>> On 18/02/2021 14:30, Numan Siddique wrote:
>>> On Wed, Feb 17, 2021 at 7:37 PM Mark Gray wrote:
Thanks Numan, some suggestions below!
>>>
>>> Hi Mark G,
>>>
>>> Thanks for the review.
On Thu, Feb 18, 2021 at 10:57 PM Mark Gray wrote:
>
> On 18/02/2021 17:21, Numan Siddique wrote:
> > On Thu, Feb 18, 2021 at 9:08 PM Mark Gray wrote:
> >>
> >> On 18/02/2021 14:30, num...@ovn.org wrote:
> >>> From: Numan Siddique
> >>>
> >>> When a Gateway router is configured with a load
On Thu, Feb 18, 2021 at 10:34 AM Dumitru Ceara wrote:
>
> On 2/18/21 9:09 AM, Frode Nordahl wrote:
> > Hello all,
>
> Hi Frode,
>
> Many thanks for this!
You're very welcome, thank you for all the bits!
> >
> > In Ubuntu we have backported a set of fixes to our OVN 20.06 and OVN
> > 20.03
On 18/02/2021 17:21, Numan Siddique wrote:
> On Thu, Feb 18, 2021 at 9:08 PM Mark Gray wrote:
>>
>> On 18/02/2021 14:30, num...@ovn.org wrote:
>>> From: Numan Siddique
>>>
>>> When a Gateway router is configured with a load balancer
>>> and it is also configured with options:lb_force_snat_ip=,
On Thu, Feb 18, 2021 at 9:15 PM Mark Gray wrote:
>
> On 18/02/2021 14:30, Numan Siddique wrote:
> > On Wed, Feb 17, 2021 at 7:37 PM Mark Gray wrote:
> >>
> >> Thanks Numan, some suggestions below!
> >
> > Hi Mark G,
> >
> > Thanks for the review.
> >
> > PSB for a few comments.
> >
> >>
> >> On
On Thu, Feb 18, 2021 at 9:08 PM Mark Gray wrote:
>
> On 18/02/2021 14:30, num...@ovn.org wrote:
> > From: Numan Siddique
> >
> > When a Gateway router is configured with a load balancer
> > and it is also configured with options:lb_force_snat_ip=,
> > OVN after load balancing the destination IP
On 18 Feb 2021, at 16:24, William Tu wrote:
> Fix the example use case based on the decription.
> EIR and CIR are measured in bytes/sec and considered 64-byte
> IP packets size withtout 14-byte Ethernet header.
> So fix the 1000pps example by: (64 - 14) * 1000 = 50,000
> If the frame includes
The prerequisite for binding_seqno_run() is that it has to be called
while there are valid SB and OVS transactions available.
ovn-controller's main() function respects that but that means that there
is a small window (when OVS transaction is still in progress) when port
bindings might be deleted
<...>
> Sending to Marko. As he wasn't subscribed to ovs-dev then.
>
<...>
> > VXLAN decap in OVS-DPDK configuration consists of two flows:
> > F1: in_port(ens1f0),eth(),ipv4(),udp(), actions:tnl_pop(vxlan_sys_4789)
> > F2: tunnel(),in_port(vxlan_sys_4789),eth(),ipv4(), actions:ens1f0_0
> >
> >
From: Numan Siddique
Presently, the 'flow_output' engine node recomputes
physical flows by calling physical_run() in the
'physical_flow_changes' handler.in some scenarios.
Because of this, an engine run can do a full recompute
of physical flows but not full recompute of logical flows.
Although
On 18/02/2021 14:30, Numan Siddique wrote:
> On Wed, Feb 17, 2021 at 7:37 PM Mark Gray wrote:
>>
>> Thanks Numan, some suggestions below!
>
> Hi Mark G,
>
> Thanks for the review.
>
> PSB for a few comments.
>
>>
>> On 09/02/2021 18:44, num...@ovn.org wrote:
>>> From: Numan Siddique
>>>
>>>
On 18/02/2021 14:30, num...@ovn.org wrote:
> From: Numan Siddique
>
> When a Gateway router is configured with a load balancer
> and it is also configured with options:lb_force_snat_ip=,
> OVN after load balancing the destination IP to one of the
> backend also does a NAT on the source ip with
Fix the example use case based on the decription.
EIR and CIR are measured in bytes/sec and considered 64-byte
IP packets size withtout 14-byte Ethernet header.
So fix the 1000pps example by: (64 - 14) * 1000 = 50,000
If the frame includes 4-byte FCS header, then it's
(64 - 14 - 4) * 1000 = 46,000
On Thu, Feb 18, 2021 at 6:55 AM Eelco Chaudron wrote:
>
>
>
> On 18 Feb 2021, at 15:40, William Tu wrote:
>
> > On Thu, Feb 18, 2021 at 1:23 AM Eelco Chaudron
> > wrote:
> >>
> >>
> >>
> >> On 17 Feb 2021, at 18:41, William Tu wrote:
> >>
>
> >> Information Rate). High priority traffic
On Thu, Feb 18, 2021 at 8:31 PM 0-day Robot wrote:
>
> 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 94
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 94 characters long (recommended limit is 79)
WARNING: Line has trailing
On 18 Feb 2021, at 15:40, William Tu wrote:
On Thu, Feb 18, 2021 at 1:23 AM Eelco Chaudron
wrote:
On 17 Feb 2021, at 18:41, William Tu wrote:
Information Rate). High priority traffic is routed to queue 10,
which marks
all traffic as CIR, i.e. Green. All low priority traffic, queue
On Thu, Feb 18, 2021 at 1:23 AM Eelco Chaudron wrote:
>
>
>
> On 17 Feb 2021, at 18:41, William Tu wrote:
>
> >>
> Information Rate). High priority traffic is routed to queue 10,
> which marks
> all traffic as CIR, i.e. Green. All low priority traffic, queue 20,
> is
>
On Wed, Feb 17, 2021 at 7:37 PM Mark Gray wrote:
>
> Thanks Numan, some suggestions below!
Hi Mark G,
Thanks for the review.
PSB for a few comments.
>
> On 09/02/2021 18:44, num...@ovn.org wrote:
> > From: Numan Siddique
> >
> > When a Gateway router is configured with a load balancer
> >
From: Numan Siddique
When a Gateway router is configured with a load balancer
and it is also configured with options:lb_force_snat_ip=,
OVN after load balancing the destination IP to one of the
backend also does a NAT on the source ip with the
lb_force_snat_ip if the packet is destined to a load
Ilya Maximets writes:
> 'distcheck' target builds the whole OVS, but for some reason we're
> running it in a single-threaded mode. Running in parallel saves
> a couple of minutes.
>
> Signed-off-by: Ilya Maximets
> ---
Cool - from what I can tell of the jobs, it shaves a couple minutes off
Sending to Marko. As he wasn't subscribed to ovs-dev then.
> -Original Message-
> From: dev On Behalf Of Eli Britstein
> Sent: Wednesday, February 10, 2021 8:57 PM
> To: d...@openvswitch.org; Ilya Maximets
> Cc: Eli Britstein ; Ameer Mahagneh
> ; Majd Dibbiny ; Gaetan Rivet
>
>
> On 17/02/2021 19:33, Lorenzo Bianconi wrote:
>
> Should the subject have an underscore? 'coverage_counters'
ops, sorry..I forgot to fix it.
>
> > In order to help understanding system behaviour for debugging purpose,
> > introduce coverage counters for run handlers of ovn-controller
> > I-P
On 2/18/21 1:20 PM, Tonghao Zhang wrote:
> On Thu, Feb 18, 2021 at 6:33 PM Ilya Maximets wrote:
>>
>> 'distcheck' target builds the whole OVS, but for some reason we're
>> running it in a single-threaded mode. Running in parallel saves
>> a couple of minutes.
>>
>> Signed-off-by: Ilya Maximets
On Thu, Feb 18, 2021 at 6:33 PM Ilya Maximets wrote:
>
> 'distcheck' target builds the whole OVS, but for some reason we're
> running it in a single-threaded mode. Running in parallel saves
> a couple of minutes.
>
> Signed-off-by: Ilya Maximets
> ---
> .ci/linux-build.sh | 2 +-
> 1 file
On 17/02/2021 19:33, Lorenzo Bianconi wrote:
Should the subject have an underscore? 'coverage_counters'
> In order to help understanding system behaviour for debugging purpose,
> introduce coverage counters for run handlers of ovn-controller
> I-P engine nodes. Moreover add a global counter for
These are some initial unit tests for PMD auto load balance that focus
on enabling/disabling and user parameters.
While writing these, I found some gaps in the logs and sent patches [1].
The existing logs were not incorrect but there was a few scenarios where
it would be reasonable and helpful to
These tests focus on enabling/disabling and user parameters.
Signed-off-by: Kevin Traynor
---
tests/alb.at | 256 +
tests/automake.mk | 1 +
tests/testsuite.at | 1 +
3 files changed, 258 insertions(+)
create mode 100644 tests/alb.at
diff
'distcheck' target builds the whole OVS, but for some reason we're
running it in a single-threaded mode. Running in parallel saves
a couple of minutes.
Signed-off-by: Ilya Maximets
---
.ci/linux-build.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/.ci/linux-build.sh
Bleep bloop. Greetings 贺鹏, 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 Peng He needs to sign off.
WARNING: Unexpected sign-offs from developers who are not
Hi, I've already sent a v2 patch.
Mark Gray 于2021年2月17日周三 下午6:44写道:
>
> On 17/02/2021 10:40, 贺鹏 wrote:
> > Hi,
> >
> > Thanks for the review.
> >
> > Mark Gray 于2021年2月17日周三 下午6:13写道:
> >>
> >> I'm not too familiar with this code but I have some comments.
> >>
> >> On 15/02/2021 09:50, Peng He
CT zone could be set from a field that is not included in frozen
metadata. Consider the example rules which are typically seen in
OpenStack security group rules:
priority=100,in_port=1,tcp,ct_state=-trk,action=ct(zone=5,table=0)
On 2/18/21 9:09 AM, Frode Nordahl wrote:
Hello all,
Hi Frode,
Many thanks for this!
In Ubuntu we have backported a set of fixes to our OVN 20.06 and OVN
20.03 packages. In preparing for new point releases for OVN 20.06
and OVN 20.03, we would very much like to add these patches to the
On 17 Feb 2021, at 18:41, William Tu wrote:
>>
Information Rate). High priority traffic is routed to queue 10,
which marks
all traffic as CIR, i.e. Green. All low priority traffic, queue 20,
is
marked as EIR, i.e. Yellow::
$ ovs-vsctl --timeout=5 set
Bleep bloop. Greetings Frode Nordahl, 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: Unexpected sign-offs from developers who are not authors or co-authors
or
Bleep bloop. Greetings Frode Nordahl, 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: patch fragment without header at line 127: @@ -861,6 +955,7 @@
From: Dumitru Ceara
The test was checking the output of "ovs-ofctl dump-flows" without
taking into account that some fields don't have predictable values,
e.g., hard_age.
Instead, use ofctl_strip_all().
Fixes: 986b3d5e4ad6 ("ofctrl.c: Add a predictable resolution for conflicting
flow
From: Han Zhou
In a scenario in I-P when a desired flow is removed and an exactly same flow is
added back in the same iteration, the merge_tracked_flows() function will merge
the operation without firing any OpenFlow action. However, if there are
multiple desired flows (e.g. F1 and F2) matching
From: Dumitru Ceara
Extend the already existing ofctl_strip() to also ignore n_packets,
n_bytes, cookie. Use some helper functions from
tests/system-userspace-packet-type-aware.at, which will be removed
by a future commit.
Signed-off-by: Dumitru Ceara
Signed-off-by: Numan Siddique
(cherry
From: Dumitru Ceara
Fix some of the race conditions that are present in the current OVN test
suite.
Signed-off-by: Dumitru Ceara
Signed-off-by: Ben Pfaff
(cherry picked from commit d6594a4695084e1b0f4fb3170547942876d3a81a)
Signed-off-by: Frode Nordahl
---
tests/ovn.at | 48
From: Han Zhou
With incremental processing for flow computation, the bottle neck of
ovn-controller in large scale environment is in the flow installation
(ofctrl_put()), which does full comparison between the two big flow tables: the
installed flows and desired flows.
This patch implements
From: Dumitru Ceara
Until now, in case the ACL configuration generates openflows that have
the same match but different actions, ovn-controller was using the
following approach:
1. If the flow being added contains conjunctive actions, merge its
actions with the already existing flow.
2.
From: Dumitru Ceara
Always consider the first "desired flow" in the list of flows that refer an
"installed flow" as the active flow. This simplifies the logic of the flow
update code and is used in a further commit to implement a partial ordering
of desired flows within installed flows.
From: Dumitru Ceara
Fixes: 6f0b1e02d9ab ("ofctrl: Incremental processing for flow installation by
tracking.")
Signed-off-by: Dumitru Ceara
Signed-off-by: Han Zhou
(cherry picked from commit 33c15c145988daa6172928dc870f3a0225515f50)
Signed-off-by: Frode Nordahl
---
controller/ofctrl.c | 4
From: Han Zhou
In update_installed_flows_by_compare() there are two loops. The first
loop iterates the installed flows and find its peer in desired flows to:
1. uninstall flows that are not needed anymore
2. update flows if needed
At the same time, it links the desired flow found for the
From: Lorenzo Bianconi
This reverts commit c0bf32d72f8b893bbe3cb64912b0fd259d71555f.
Signed-off-by: Lorenzo Bianconi
Signed-off-by: Han Zhou
(cherry picked from commit d9ed450713eda62af1bec5009694b2d206c9f435)
Signed-off-by: Frode Nordahl
---
northd/ovn-northd.8.xml | 37
From: Dumitru Ceara
In ofctrl_add_or_append_flow() when merging flow actions make sure we only
do that for conjunctive flows. All other actions can not be merged with
action "conjunction".
CC: Mark Michelson
Fixes: e659bab31a91 ("Combine conjunctions with identical matches into one
flow.")
From: Dumitru Ceara
Instead of removing the old desired flow from the list and inserting the new
one at the end we now directly replace the old flow with the new one while
maintaining the same ordering.
For now order of the flows is not relevant but further commits require
maintaining the order
From: Numan Siddique
when ovn-ctl run_(nb_sb)_ovsdb is called, the ovsdb-server is started without
passing --detach and --monoitor and the process is exec'd.
For cluster mode, upgrade_cluster is never called and hence the dbs are not
upraded
to new schema. CMS has to handle the db upgrade
From: Han Zhou
This patch optimizes the previous patch that incrementally processes
flow installation by merging the "add-after-delete" flow changes as
much as possible to avoid unnecessary OpenFlow updates.
Acked-by: Mark Michelson
Signed-off-by: Han Zhou
(cherry picked from commit
From: Han Zhou
Currently there is no link maintained between installed flows and desired
flows. This patch maintains the mapping between them, which will be useful
for a future patch that incrementally processes the flow installation without
having to do the full comparison between them.
This
From: Han Zhou
These functions will be reused in multiple places in a future patch.
Acked-by: Mark Michelson
Signed-off-by: Han Zhou
(cherry picked from commit 23063cf4178c05f5d6b3e4ec6d323ccc88df6101)
Signed-off-by: Frode Nordahl
---
controller/ofctrl.c | 103
Hello all,
In Ubuntu we have backported a set of fixes to our OVN 20.06 and OVN
20.03 packages. In preparing for new point releases for OVN 20.06
and OVN 20.03, we would very much like to add these patches to the
upstream branches.
The issues solved for OVN 20.03 with these patches are:
*
The test was checking the output of "ovs-ofctl dump-flows" without
taking into account that some fields don't have predictable values,
e.g., hard_age.
Instead, use ofctl_strip_all().
Fixes: 986b3d5e4ad6 ("ofctrl.c: Add a predictable resolution for conflicting
flow actions.")
Reported-by: Numan
Extend the already existing ofctl_strip() to also ignore n_packets,
n_bytes, cookie. Use some helper functions from
tests/system-userspace-packet-type-aware.at, which will be removed
by a future commit.
Signed-off-by: Dumitru Ceara
Signed-off-by: Numan Siddique
(cherry picked from commit
Fix some of the race conditions that are present in the current OVN test
suite.
Signed-off-by: Dumitru Ceara
Signed-off-by: Ben Pfaff
(cherry picked from commit d6594a4695084e1b0f4fb3170547942876d3a81a)
Signed-off-by: Frode Nordahl
---
tests/ovn.at | 48
Until now, in case the ACL configuration generates openflows that have
the same match but different actions, ovn-controller was using the
following approach:
1. If the flow being added contains conjunctive actions, merge its
actions with the already existing flow.
2. Otherwise, if the flow is
Fixes: 6f0b1e02d9ab ("ofctrl: Incremental processing for flow installation by
tracking.")
Signed-off-by: Dumitru Ceara
Signed-off-by: Han Zhou
(cherry picked from commit 33c15c145988daa6172928dc870f3a0225515f50)
Signed-off-by: Frode Nordahl
---
controller/ofctrl.c | 4 ++--
1 file changed, 2
Always consider the first "desired flow" in the list of flows that refer an
"installed flow" as the active flow. This simplifies the logic of the flow
update code and is used in a further commit to implement a partial ordering
of desired flows within installed flows.
Signed-off-by: Dumitru Ceara
Instead of removing the old desired flow from the list and inserting the new
one at the end we now directly replace the old flow with the new one while
maintaining the same ordering.
For now order of the flows is not relevant but further commits require
maintaining the order of desired flows.
In ofctrl_add_or_append_flow() when merging flow actions make sure we only
do that for conjunctive flows. All other actions can not be merged with
action "conjunction".
CC: Mark Michelson
Fixes: e659bab31a91 ("Combine conjunctions with identical matches into one
flow.")
Signed-off-by: Dumitru
Hi guys,
On 2/18/21 12:16 AM, Han Zhou wrote:
On Wed, Feb 17, 2021 at 2:36 PM Tim Rozet wrote:
On Tue, Feb 16, 2021 at 4:21 AM Han Zhou wrote:
On Wed, Feb 10, 2021 at 2:17 PM Lorenzo Bianconi <
lorenzo.bianc...@redhat.com> wrote:
Increase priority for automatic routes (routes
In a scenario in I-P when a desired flow is removed and an exactly same flow is
added back in the same iteration, the merge_tracked_flows() function will merge
the operation without firing any OpenFlow action. However, if there are
multiple desired flows (e.g. F1 and F2) matching the same
In update_installed_flows_by_compare() there are two loops. The first
loop iterates the installed flows and find its peer in desired flows to:
1. uninstall flows that are not needed anymore
2. update flows if needed
At the same time, it links the desired flow found for the installed flow
which
This patch optimizes the previous patch that incrementally processes
flow installation by merging the "add-after-delete" flow changes as
much as possible to avoid unnecessary OpenFlow updates.
Acked-by: Mark Michelson
Signed-off-by: Han Zhou
(cherry picked from commit
With incremental processing for flow computation, the bottle neck of
ovn-controller in large scale environment is in the flow installation
(ofctrl_put()), which does full comparison between the two big flow tables: the
installed flows and desired flows.
This patch implements tracking desired flow
These functions will be reused in multiple places in a future patch.
Acked-by: Mark Michelson
Signed-off-by: Han Zhou
(cherry picked from commit 23063cf4178c05f5d6b3e4ec6d323ccc88df6101)
Signed-off-by: Frode Nordahl
---
controller/ofctrl.c | 103 ++--
1
Currently there is no link maintained between installed flows and desired
flows. This patch maintains the mapping between them, which will be useful
for a future patch that incrementally processes the flow installation without
having to do the full comparison between them.
This patch also
when ovn-ctl run_(nb_sb)_ovsdb is called, the ovsdb-server is started without
passing --detach and --monoitor and the process is exec'd.
For cluster mode, upgrade_cluster is never called and hence the dbs are not
upraded
to new schema. CMS has to handle the db upgrade separately.
This patch
Hello all,
In Ubuntu we have backported a set of fixes to our OVN 20.06 and OVN
20.03 packages. In preparing for new point releases for OVN 20.06
and OVN 20.03, we would very much like to add these patches to the
upstream branches.
The issues solved for OVN 20.06 with these patches are:
*
94 matches
Mail list logo