Re: [ovs-dev] [PATCH ovn v5 0/6] northd: Introduce incremental processing framework

Thanks Han! On Wed 17 Nov 2021, 06:58 Han Zhou, wrote: > > > On Tue, Nov 9, 2021 at 11:36 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 configura

Re: [ovs-dev] ovs-vswitchd too large memory consumption with OVN stateless ACL

On 11/20/21 03:39, Han Zhou wrote: > > > On Fri, Nov 19, 2021 at 3:11 PM Ilya Maximets > wrote: >> >> On 11/19/21 19:12, Vladislav Odintsov wrote: >> > Hi, >> > >> > I’m testing OVN stateless ACL rules with `$port_group_ipVERSION` in match >> > portion. >> > There’s a

Re: [ovs-dev] ovs-vswitchd too large memory consumption with OVN stateless ACL

On Fri, Nov 19, 2021 at 3:11 PM Ilya Maximets wrote: > > On 11/19/21 19:12, Vladislav Odintsov wrote: > > Hi, > > > > I’m testing OVN stateless ACL rules with `$port_group_ipVERSION` in match portion. > > There’s a strange behaviour and sometimes I got configuration, which totally kills my transpo

Re: [ovs-dev] [PATCH ovn v3 2/2] northd: Use ct_(snat/dnat)_in_czone action for distributed routers.

On Fri, Nov 19, 2021 at 1:11 PM Mark Michelson wrote: > > Thanks for the update, Numan. > > Acked-by: Mark Michelson Thanks. I applied both the patches to the main branch. Numan > > On 11/19/21 11:32, num...@ovn.org wrote: > > From: Numan Siddique > > > > Make of use of these new actions for

Re: [ovs-dev] ovs-vswitchd too large memory consumption with OVN stateless ACL

On 11/19/21 19:12, Vladislav Odintsov wrote: > Hi, > > I’m testing OVN stateless ACL rules with `$port_group_ipVERSION` in match > portion. > There’s a strange behaviour and sometimes I got configuration, which totally > kills my transport nodes, where logical switch ports reside. > ovs-vswitchd

[ovs-dev] [PATCH] ofproto: Fix resource usage explosion while processing bundled FLOW_MOD.

While processing a bundle, OVS will add all new and modified rules to classifiers. Classifiers are using RCU-protected pvector to store subtables. Addition of a new subtable or removal of the old one leads to re-allocation and memory copy of the pvector array. Old version of that array is given t

Re: [ovs-dev] [PATCH ovn v3 0/4] uuid-based conjunction id generation.

On Fri, Nov 19, 2021 at 10:55 AM Numan Siddique wrote: > > On Thu, Nov 11, 2021 at 5:38 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 lfl

Re: [ovs-dev] [PATCH ovn v3 4/4] lflow: Consistent conjunction id generation.

On Fri, Nov 19, 2021 at 11:13 AM Numan Siddique wrote: > > On Thu, Nov 11, 2021 at 5:39 PM Han Zhou wrote: > > > > 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 unne

Re: [ovs-dev] [PATCH ovn v3 4/4] lflow: Consistent conjunction id generation.

On Thu, Nov 11, 2021 at 5:39 PM Han Zhou wrote: > > 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 probl

Re: [ovs-dev] [PATCH ovn v3 0/4] uuid-based conjunction id generation.

On Thu, Nov 11, 2021 at 5:38 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 n_conj

[ovs-dev] ovs-vswitchd too large memory consumption with OVN stateless ACL

Hi, I’m testing OVN stateless ACL rules with `$port_group_ipVERSION` in match portion. There’s a strange behaviour and sometimes I got configuration, which totally kills my transport nodes, where logical switch ports reside. ovs-vswitchd and ovn-controller processes utilise 100% 1 core CPU each

Re: [ovs-dev] [PATCH ovn v3 2/2] northd: Use ct_(snat/dnat)_in_czone action for distributed routers.

Thanks for the update, Numan. Acked-by: Mark Michelson On 11/19/21 11:32, num...@ovn.org wrote: From: Numan Siddique Make of use of these new actions for the distributed routers for NAT. These new actions ensure that both sNAT and dNAT happens in the same zone. This approach solves a coupl

Re: [ovs-dev] Flaky ALB tests.

On 19/11/2021 17:37, Ilya Maximets wrote: On 11/19/21 18:09, Kevin Traynor wrote: On 19/11/2021 12:34, Kevin Traynor wrote: On 19/11/2021 12:25, Ilya Maximets wrote: Hi, Kevin. Hi Ilya, After the recent ALB commits, some of these tests are failing more or less frequently in our CI and on

Re: [ovs-dev] [PATCH] ovsdb-idl: Re-parse backrefs of inserted rows only once.

On 11/19/21 12:02, Dumitru Ceara wrote: > Hi Ilya, > > On 11/19/21 3:22 AM, Ilya Maximets wrote: >> While adding new rows ovsdb-idl re-parses all the other rows that >> references this new one. For example, current ovn-kubernetes creates >> load balancers and adds the same load balancer to all lo

Re: [ovs-dev] [PATCH] ofproto-dpif-xlate: Terminate native tunnels only on ports with IP addresses.

On 11/17/21 15:22, Mike Pattrick wrote: > Looks good to me. > > Acked-by: Mike Pattrick Thanks! I folded the test change in and applied. Best regards, Ilya Maximets. > > > On Mon, Nov 15, 2021 at 1:08 PM Ilya Maximets wrote: >> >> On 11/15/21 18:28, Mike Pattrick wrote: >>> Hello Ilya, >>>

Re: [ovs-dev] Flaky ALB tests.

On 11/19/21 18:09, Kevin Traynor wrote: > On 19/11/2021 12:34, Kevin Traynor wrote: >> On 19/11/2021 12:25, Ilya Maximets wrote: >>> Hi, Kevin. >>> >> >> Hi Ilya, >> >>> After the recent ALB commits, some of these tests are failing more >>> or less frequently in our CI and on my local setup too.  C

Re: [ovs-dev] Flaky ALB tests.

On 19/11/2021 12:34, Kevin Traynor wrote: On 19/11/2021 12:25, Ilya Maximets wrote: Hi, Kevin. Hi Ilya, After the recent ALB commits, some of these tests are failing more or less frequently in our CI and on my local setup too. Could you, please, take a look? Sure, I'll take a look. Than

Re: [ovs-dev] [PATCH ovn v2 2/2] northd: Use ct_(snat/dnat)_in_czone action for distributed routers.

On Fri, Nov 19, 2021 at 11:16 AM Mark Michelson wrote: > > On 11/18/21 18:42, Numan Siddique wrote: > > On Thu, Nov 18, 2021 at 4:49 PM Mark Michelson wrote: > >> > >> Hi Numan, > >> > >> I had one question in the code and then one small finding in the > >> documentation. See below: > >> > >> On

[ovs-dev] [PATCH ovn v3 2/2] northd: Use ct_(snat/dnat)_in_czone action for distributed routers.

From: Numan Siddique Make of use of these new actions for the distributed routers for NAT. These new actions ensure that both sNAT and dNAT happens in the same zone. This approach solves a couple of problems: - The datapath flows generated for external traffic which requires dNAT (N -> S)

[ovs-dev] [PATCH ovn v3 1/2] actions: Add new actions - ct_dnat_in_czone and ct_snat_in_czone.

From: Numan Siddique These actions are very similar to ct_dnat and ct_snat respectively with one difference. These new actions use the same zone id for natting. Upcoming patch will make use of these actions. Acked-by: Mark Michelson Signed-off-by: Numan Siddique --- v2 -> v3 * Added M

Re: [ovs-dev] [PATCH ovn v9 5/6] ic: add support for routing tables in adv/learn routes

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 is 81 characters long (recommended limit is 79) #409 FILE: ic/ovn-ic.c:1352

Re: [ovs-dev] [PATCH ovn v9 4/6] northd, utils: support for RouteTables in LRs

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 lacks whitespace around operator WARNING: Line lacks whitespace around oper

Re: [ovs-dev] [PATCH] netlink-socket: Check for null sock in nl_sock_recv__()

On 11/18/21 22:19, David Christensen wrote: > > > On 11/18/21 11:56 AM, Murilo Opsfelder Araújo wrote: >> On 11/16/21 19:31, Ilya Maximets wrote: >>> On 10/25/21 19:45, David Christensen wrote: In certain high load situations, such as when creating a large number of ports on a switch, t

Re: [ovs-dev] [PATCH ovn v2 2/2] northd: Use ct_(snat/dnat)_in_czone action for distributed routers.

On 11/18/21 18:42, Numan Siddique wrote: On Thu, Nov 18, 2021 at 4:49 PM Mark Michelson wrote: Hi Numan, I had one question in the code and then one small finding in the documentation. See below: On 11/18/21 13:13, num...@ovn.org wrote: From: Numan Siddique Make of use of these new action

Re: [ovs-dev] [PATCH ovn v8 2/5] northd: make connected routes have higher priority than static

Done: https://patchwork.ozlabs.org/project/ovn/patch/2029160721.60612-2-odiv...@gmail.com/ Regards, Vladislav Odintsov > On 19 Nov 2021, at 17:55, Numan Siddique wrote: > > On Fri, Nov 19, 2021 at 8:12 AM Vladislav Odintsov > wrote: >> >> Hi Numan, >> >> yes, i

[ovs-dev] [PATCH ovn v9 5/6] ic: add support for routing tables in adv/learn routes

Previously support for multiple routing tables was added to northd code. This commit expands support for multiple routing tables by adding support of advertising and learning routes with their routing table information. To utilize such feature, user must: 1. create Logical Router in each AZ; 2. cr

[ovs-dev] [PATCH ovn v9 4/6] northd, utils: support for RouteTables in LRs

This patch extends Logical Router's routing functionality. Now user may create multiple routing tables within a Logical Router and assign them to Logical Router Ports. Traffic coming from Logical Router Port with assigned route_table is checked against Logical_Router_Static_Routes with same route_

[ovs-dev] [PATCH ovn v9 6/6] ic: don't learn routes which have local GW

In case we have ovn-ic-interconnected Logical_Routers and install same ip_prefix route with GW in local AZ in each LR in each AZ, this route would be learned in other AZs and L3 loop is possible. There could be next routes output: [az1 ~]$ ovn-nbctl lr-route-list lr0 IPv4 Routes Route Table global

[ovs-dev] [PATCH ovn v9 3/6] northd: make connected routes have higher priority than static

With this patch routes to connected networks have higher priority than static routes with same ip_prefix. This brings commonly-used behaviour for routes lookup order: 1: longest prefix match 2: metric The metric has next lookup order: 1: connected routes 2: static routes Earlier static and conne

[ovs-dev] [PATCH ovn v9 2/6] ic: maintain route origin - connected/static

This patch adds ability to save route's origin while IC learning. Directly connected routes are saved in IC SB DB with "connected" origin column value. Static routes have "static" value in origin column. This logic would be used in next patch to compute priority for lr_in_ip_routing stage lflows.

[ovs-dev] [PATCH ovn v9 1/6] northd: fix symmetric ECMP reply flows documentation

This patch fixes a flow priority which is added in IP Defrag table and adds priority value for symmetric ECMP reply flow in IP Routing table. Fixes: 384a7c6 (northd: Refactor Logical Flows for routers with DNAT/Load Balancers) Signed-off-by: Vladislav Odintsov --- northd/ovn-northd.8.xml | 18 +

[ovs-dev] [PATCH ovn v9 0/6] Add multiple routing tables support to Logical Routers

v8 -> v9: - Fix documentation errors. - Fix commit messages. v7 -> v8: - Updated manpages with and commit messages according to latest changes. - Small fixes. v6 -> v7: - Addressed Han's comments regarding saving same behaviour for route-tables static routes as for static routes fro

[ovs-dev] [PATCH ovn 3/3] inc-proc-eng: Add inc-engine/recompute command.

This will be automatically available to all incremental processing engine clients (ovn-northd and ovn-controller). At the same time, deprecate the ovn-controller "recompute" command. Signed-off-by: Dumitru Ceara --- NEWS|2 ++ controller/ovn-controller.c |7 +++--

[ovs-dev] [PATCH ovn 2/3] ovn-northd: Don't trigger recompute for transactions that are in progress.

Most of the times ovsdb_idl_loop_commit_and_wait() will return -1 indicating that the newly initiated transaction is still in progress. That's not an error condition and it shouldn't trigger a forced recompute. Moreover, arguably, we shouldn't clear tracked rows only when the transaction is succe

[ovs-dev] [PATCH ovn 1/3] inc-proc-northd: Don't reset I-P engine force recompute.

In some cases, e.g., when ovn-northd reconnects to NB or SB, a full recompute must be triggered. In such cases inc_proc_northd_run() is called with 'recompute=true' and it should in turn call engine_set_force_recompute(true). However, when 'recompute=false', inc_proc_northd_run() should not reset

[ovs-dev] [PATCH ovn 0/3] Reduce number of recomputes in northd.

Patches 1/3 and 2/3 fix some issues with the way the incremental processing engine is currently used for northd. Patch 3/3 enables the "recompute" command for northd too. Dumitru Ceara (3): inc-proc-northd: Don't reset I-P engine force recompute. ovn-northd: Don't trigger recompute fo

Re: [ovs-dev] [PATCH 2/2] system-dpdk: Use dummy-pmd port for packet injection.

On 11/18/21 10:16, Maxime Coquelin wrote: > Hi David, > > On 9/27/21 15:57, David Marchand wrote: >> net_pcap is not always available in DPDK (like, in a dev >> environment when you forgot to install the libpcap-devel). >> On the other hand, OVS already has its own way to inject packets into a >>

Re: [ovs-dev] [PATCH] dpif-netdev: Call cpuid for x86 isa availability.

On 11/19/21 16:04, David Marchand wrote: > On Wed, Nov 3, 2021 at 4:32 PM Ilya Maximets wrote: >> >> 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 is

Re: [ovs-dev] [PATCH] dpif-netdev: Call cpuid for x86 isa availability.

On Wed, Nov 3, 2021 at 4:32 PM Ilya Maximets wrote: > > 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 chan

Re: [ovs-dev] [PATCH ovn v8 2/5] northd: make connected routes have higher priority than static

On Fri, Nov 19, 2021 at 8:12 AM Vladislav Odintsov wrote: > > Hi Numan, > > yes, it’s ready. But I’ve based it on the commit from this patch [1]. > Can you please take a look on it and apply if it’s okay. > Then I can send patch series without conflicting changes. > > Or I can send it with that pa

[ovs-dev] [PATCH v4] utilities: Add another GDB macro for ovs-vswitchd

This commit adds a basic packet metadata macro to the already existing macros in ovs_gdb.py, ovs_dump_packets will print out information about one or more packets. It feeds packets into tcpdump, and the user can pass in tcpdump options to modify how packets are parsed or even write out packets to a

Re: [ovs-dev] [PATCH dpdk-latest v3 0/5] Fixes for DPDK v21.11

Hello Ian, On Wed, Nov 10, 2021 at 3:19 PM Stokes, Ian wrote: > Thanks for the series. Looks good on the github actions to myself. I'd like > to get OVS up and running on validation with rc2 so I've pushed the series to > dpdk-latest and rebased. > > Folks on the Intel side will be carrying out

Re: [ovs-dev] [PATCH v3] netdev-linux: Ingress policing to use matchall if basic is, not available.

On 12 Nov 2021, at 18:49, Mike Pattrick wrote: > Currently ingress policing uses the basic classifier to apply traffic > control filters if hardware offload is not enabled, else it uses > matchall. This change enables fallback onto the matchall classifier for > cases when the kernel is not buil

Re: [ovs-dev] [PATCH v1] checkpatch: Correct line count in error messages

Mike Pattrick writes: > As part of some previous checkpatch work, we discovered that checkpatch > isn't always reporting correct line numbers. As it turns out, Python's > splitlines function considers several characters to be new lines which > common text editors do not typically consider to be n

Re: [ovs-dev] [PATCH] doc: fix install doc

On 5 Nov 2021, at 6:44, lic121 wrote: > remove bad quotes > > Signed-off-by: lic121 Thanks for the patch, looks good to me. Acked-by: Eelco Chaudron ___ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Re: [ovs-dev] [PATCH ovn v8 2/5] northd: make connected routes have higher priority than static

Hi Numan, yes, it’s ready. But I’ve based it on the commit from this patch [1]. Can you please take a look on it and apply if it’s okay. Then I can send patch series without conflicting changes. Or I can send it with that patch as a part of patchset. 1: https://patchwork.ozlabs.org/project/ovn/

Re: [ovs-dev] [PATCH v4] utilities: Add another GDB macro for ovs-vswitchd

Hi Mike, I tried to apply this patch, but for some reason, it’s messed up, guess some longer lines are truncated? Patchwork also does not understand it. Maybe you can check and send a v5? Cheers, Eelco On 12 Nov 2021, at 18:38, Mike Pattrick wrote: > This commit adds a basic packet metadat

Re: [ovs-dev] [RFC PATCH] netdev-dpdk: Expose per rxq/txq basic statistics.

On 19/11/2021 12:45, Maxime Coquelin wrote: On 11/19/21 13:08, Kevin Traynor wrote: On 19/11/2021 10:31, David Marchand wrote: On Fri, Nov 19, 2021 at 10:40 AM Maxime Coquelin wrote: Hi Kevin, David, On 11/18/21 16:31, Kevin Traynor wrote: On 15/10/2021 16:04, David Marchand wrote: When

Re: [ovs-dev] [RFC PATCH] netdev-dpdk: Expose per rxq/txq basic statistics.

On 11/19/21 13:08, Kevin Traynor wrote: On 19/11/2021 10:31, David Marchand wrote: On Fri, Nov 19, 2021 at 10:40 AM Maxime Coquelin wrote: Hi Kevin, David, On 11/18/21 16:31, Kevin Traynor wrote: On 15/10/2021 16:04, David Marchand wrote: When troubleshooting multiqueue setups, having per

Re: [ovs-dev] Flaky ALB tests.

On 19/11/2021 12:25, Ilya Maximets wrote: Hi, Kevin. Hi Ilya, After the recent ALB commits, some of these tests are failing more or less frequently in our CI and on my local setup too. Could you, please, take a look? Sure, I'll take a look. Thanks for letting me know, I hadn't seen it.

[ovs-dev] Flaky ALB tests.

Hi, Kevin. After the recent ALB commits, some of these tests are failing more or less frequently in our CI and on my local setup too. Could you, please, take a look? Below is one of the recent examples. Another test that fails with similar symptoms periodically is 'ALB - min num PMD/RxQ'. It lo

Re: [ovs-dev] [RFC PATCH] netdev-dpdk: Expose per rxq/txq basic statistics.

On 19/11/2021 10:31, David Marchand wrote: On Fri, Nov 19, 2021 at 10:40 AM Maxime Coquelin wrote: Hi Kevin, David, On 11/18/21 16:31, Kevin Traynor wrote: On 15/10/2021 16:04, David Marchand wrote: When troubleshooting multiqueue setups, having per queue statistics helps checking packets r

[ovs-dev] [PATCH ovn] binding: Log iface-id-ver mismatch.

This is useful for debugging. Signed-off-by: Dumitru Ceara --- controller/binding.c | 5 + 1 file changed, 5 insertions(+) diff --git a/controller/binding.c b/controller/binding.c index fcd319e6d..4d62b0858 100644 --- a/controller/binding.c +++ b/controller/binding.c @@ -2843,6 +2843,11 @@

[ovs-dev] [PATCH ovn v2] northd: Generate ARP responder flows only for reachable VIPs.

It's not useful to generate ARP responder flows for VIPs that are not reachable on any port of a logical router port. On scaled ovn-kubernetes deployments the VIP ARP responder Southbound address sets become quite large, wasting resources because they are never used. Stop generating ARP responder

[ovs-dev] [PATCH ovn] northd: I-P add explicit dependency on Load_Balancer_Group.

The commit that added I-P for northd missed to add this explicit dependency. This didn't really affect functionality because: a. ovn-northd monitors the whole NB database b. changes to the Load_Balancer_Group table mark rows in Logical_Switch table that refer to the updated LB groups as "update

Re: [ovs-dev] [PATCH] ovsdb-idl: Re-parse backrefs of inserted rows only once.

Hi Ilya, On 11/19/21 3:22 AM, Ilya Maximets wrote: > While adding new rows ovsdb-idl re-parses all the other rows that > references this new one. For example, current ovn-kubernetes creates > load balancers and adds the same load balancer to all logical switches > and logical routers. So, then a

Re: [ovs-dev] [RFC PATCH] netdev-dpdk: Expose per rxq/txq basic statistics.

On Fri, Nov 19, 2021 at 10:40 AM Maxime Coquelin wrote: > > Hi Kevin, David, > > On 11/18/21 16:31, Kevin Traynor wrote: > > On 15/10/2021 16:04, David Marchand wrote: > >> When troubleshooting multiqueue setups, having per queue statistics helps > >> checking packets repartition in rx and tx queu

Re: [ovs-dev] [RFC PATCH] netdev-dpdk: Expose per rxq/txq basic statistics.

Hi Kevin, David, On 11/18/21 16:31, Kevin Traynor wrote: On 15/10/2021 16:04, David Marchand wrote: When troubleshooting multiqueue setups, having per queue statistics helps checking packets repartition in rx and tx queues. Per queue statistics are exported by most DPDK drivers (with capabilit