| Message ID | 20210716114158.1895109-1-numans@ovn.org |
|---|---|
| Headers | show |
| Series | pflow_output and ct_zone engine improvements. | expand |
Hi Numan, Looking at this series, I realize that this section of the OVN code is not exactly my wheelhouse. While I don't spot anything obviously incorrect (e.g. memory leaks, failing tests), I think to properly give this the scrutiny it needs, it would require me to study the code again to ensure I understand all the ins and outs. I will do this in my free time so that I can give better reviews in the future, but for now in the interest of expediency, I will put: Reviewed-by: Mark Michelson <mmichels@redhat.com> I'm going to unassign myself from this review so that someone else can give a better review. On 7/16/21 7:41 AM, numans@ovn.org wrote: > From: Numan Siddique <numans@ovn.org> > > This patch series attempts to improve the handling of runtime_data > changes in the engine nodes - ct_zones and pflow_output. > It also attempts to improve the handling of ct_zones change in > pflow_output engine. > > > After this patch series, the amount of full recomputes of pflow_output > engine is reduced drastically. This helps in scaled environments. > > For example with the simple ovn-scale-test which creates 20 logical > ports (10 each in one logical switch) and binds it equally on two chassis, > reduces the number of physical_run() calls from around 30 to around 8. > > v1 -> v2 > ---- > * Rebased to resolve the conflct. > > Numan Siddique (5): > controller: Move 'struct local_datapath' to a separate file. > binding: Add the localport port binding in the binding_lport > information. > binding: Track the changes of container lport when it's parent > changes. > I-P: Handle runtime data changes for plow_output engine. > controller: Improve ct zone handling. > > controller/automake.mk | 4 +- > controller/binding.c | 565 ++++++++++++++---------------------- > controller/binding.h | 36 ++- > controller/ldata.c | 537 ++++++++++++++++++++++++++++++++++ > controller/ldata.h | 164 +++++++++++ > controller/lflow.c | 6 +- > controller/lflow.h | 1 + > controller/lport.c | 40 +++ > controller/lport.h | 7 +- > controller/ovn-controller.c | 251 +++++++++++----- > controller/ovn-controller.h | 42 --- > controller/patch.c | 1 + > controller/physical.c | 382 ++++++------------------ > controller/physical.h | 13 +- > controller/pinctrl.c | 1 + > lib/inc-proc-eng.h | 4 + > tests/ovn-controller.at | 35 ++- > tests/ovn.at | 2 +- > 18 files changed, 1320 insertions(+), 771 deletions(-) > create mode 100644 controller/ldata.c > create mode 100644 controller/ldata.h >
On Thu, Jul 22, 2021 at 1:37 PM Mark Michelson <mmichels@redhat.com> wrote: > > Hi Numan, > > Looking at this series, I realize that this section of the OVN code is > not exactly my wheelhouse. While I don't spot anything obviously > incorrect (e.g. memory leaks, failing tests), I think to properly give > this the scrutiny it needs, it would require me to study the code again > to ensure I understand all the ins and outs. I will do this in my free > time so that I can give better reviews in the future, but for now in the > interest of expediency, I will put: > > Reviewed-by: Mark Michelson <mmichels@redhat.com> > > I'm going to unassign myself from this review so that someone else can > give a better review. Thanks Mark. I assigned to myself, and I will try to send feedback by next Monday. Han > > On 7/16/21 7:41 AM, numans@ovn.org wrote: > > From: Numan Siddique <numans@ovn.org> > > > > This patch series attempts to improve the handling of runtime_data > > changes in the engine nodes - ct_zones and pflow_output. > > It also attempts to improve the handling of ct_zones change in > > pflow_output engine. > > > > > > After this patch series, the amount of full recomputes of pflow_output > > engine is reduced drastically. This helps in scaled environments. > > > > For example with the simple ovn-scale-test which creates 20 logical > > ports (10 each in one logical switch) and binds it equally on two chassis, > > reduces the number of physical_run() calls from around 30 to around 8. > > > > v1 -> v2 > > ---- > > * Rebased to resolve the conflct. > > > > Numan Siddique (5): > > controller: Move 'struct local_datapath' to a separate file. > > binding: Add the localport port binding in the binding_lport > > information. > > binding: Track the changes of container lport when it's parent > > changes. > > I-P: Handle runtime data changes for plow_output engine. > > controller: Improve ct zone handling. > > > > controller/automake.mk | 4 +- > > controller/binding.c | 565 ++++++++++++++---------------------- > > controller/binding.h | 36 ++- > > controller/ldata.c | 537 ++++++++++++++++++++++++++++++++++ > > controller/ldata.h | 164 +++++++++++ > > controller/lflow.c | 6 +- > > controller/lflow.h | 1 + > > controller/lport.c | 40 +++ > > controller/lport.h | 7 +- > > controller/ovn-controller.c | 251 +++++++++++----- > > controller/ovn-controller.h | 42 --- > > controller/patch.c | 1 + > > controller/physical.c | 382 ++++++------------------ > > controller/physical.h | 13 +- > > controller/pinctrl.c | 1 + > > lib/inc-proc-eng.h | 4 + > > tests/ovn-controller.at | 35 ++- > > tests/ovn.at | 2 +- > > 18 files changed, 1320 insertions(+), 771 deletions(-) > > create mode 100644 controller/ldata.c > > create mode 100644 controller/ldata.h > > > > _______________________________________________ > dev mailing list > dev@openvswitch.org > https://mail.openvswitch.org/mailman/listinfo/ovs-dev
From: Numan Siddique <numans@ovn.org> This patch series attempts to improve the handling of runtime_data changes in the engine nodes - ct_zones and pflow_output. It also attempts to improve the handling of ct_zones change in pflow_output engine. After this patch series, the amount of full recomputes of pflow_output engine is reduced drastically. This helps in scaled environments. For example with the simple ovn-scale-test which creates 20 logical ports (10 each in one logical switch) and binds it equally on two chassis, reduces the number of physical_run() calls from around 30 to around 8. v1 -> v2 ---- * Rebased to resolve the conflct. Numan Siddique (5): controller: Move 'struct local_datapath' to a separate file. binding: Add the localport port binding in the binding_lport information. binding: Track the changes of container lport when it's parent changes. I-P: Handle runtime data changes for plow_output engine. controller: Improve ct zone handling. controller/automake.mk | 4 +- controller/binding.c | 565 ++++++++++++++---------------------- controller/binding.h | 36 ++- controller/ldata.c | 537 ++++++++++++++++++++++++++++++++++ controller/ldata.h | 164 +++++++++++ controller/lflow.c | 6 +- controller/lflow.h | 1 + controller/lport.c | 40 +++ controller/lport.h | 7 +- controller/ovn-controller.c | 251 +++++++++++----- controller/ovn-controller.h | 42 --- controller/patch.c | 1 + controller/physical.c | 382 ++++++------------------ controller/physical.h | 13 +- controller/pinctrl.c | 1 + lib/inc-proc-eng.h | 4 + tests/ovn-controller.at | 35 ++- tests/ovn.at | 2 +- 18 files changed, 1320 insertions(+), 771 deletions(-) create mode 100644 controller/ldata.c create mode 100644 controller/ldata.h