> Hello Ramu
>
> Can you describe your configuration for this test failure when logical
> switch
> arp responders are skipped for logical switch "router type" ports ?
> I know the existing OVN tests (both system and non-system) pass either way.
>
> Thanks Darre
, potentially, multiple hypervisors
could independently respond.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/northd/ovn-northd.8.xml | 5 +++--
ovn/northd/ovn-northd.c | 5 -
2 files changed, 7 insertions(+), 3 deletions(-)
diff --git a/ovn/northd/ovn-northd.8.xml
-if (!strcmp(op->nbsp->type, "localnet")) {
+/* Skip arp responder if the logical switch inport is not
+ * associated with a local VIF or a l2gateway port */
+if ((strcmp(op->nbsp->type, "")) &&
+(strcmp(op->nbsp->type, "l2gateway"))) {
, potentially, multiple hypervisors
could independently respond.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/northd/ovn-northd.8.xml | 4 ++--
ovn/northd/ovn-northd.c | 4 +++-
2 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/ovn/northd/ovn-northd.8.xml
>
> Does the following diff work on your system?
>
>
> diff --git a/ovn/northd/ovn-northd.8.xml b/ovn/northd/ovn-northd.8.xml
> index 7797417..4f1cd89 100644
> --- a/ovn/northd/ovn-northd.8.xml
> +++ b/ovn/northd/ovn-northd.8.xml
> @@ -421,8 +421,8 @@
>
>
>
> -Priority-100
treatment must be similar to the
arps coming in from the localnet port - which is to
skip the arp responder table.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/northd/ovn-northd.8.xml | 3 ++-
ovn/northd/ovn-northd.c | 3 ++-
2 files changed, 4 insertions(+), 2 del
Hi,
I dont see a /proc filesystem based interface to help debug the
openvswitch kernel module. Some of the knobs that may be useful are:
a) debug knobs for stats (for those stats are not already reported),
b) control knobs like: stop/start forwarding on the fast-path etc.
Such an interface would
wrote:
>
> On Fri, Aug 19, 2016 at 9:06 PM, Ramu Ramamurthy <ramu.ramamur...@gmail.com>
> wrote:
>>
>> From: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
>>
>> The following leaks are due to missing ds_destroy in a few
>> places in build_acl.
>>
&g
-string.c:142)
by 0x40D505: build_acls (ovn-northd.c:2346)
by 0x40D505: build_lswitch_flows.constprop.36 (ovn-northd.c:2472)
by 0x4072D9: build_lflows (ovn-northd.c:3845)
by 0x4072D9: ovnnb_db_run (ovn-northd.c:3971)
by 0x4072D9: main (ovn-northd.c:4375)
Signed-off-by: Ramu
-01T18:16:22Z|00018|binding|INFO|Releasing lport
eee1a9af-7513-4540-9385-9e3972bfca05 from this chassis.
2016-09-01T18:16:22Z|00019|binding|INFO|Releasing fa:16:3e:01:c3:4a 10.0.0.7
fd93:b509:aa46:0:f816:3eff:fe01:c34a
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
v1->v2:
to be at low rate normally.
Logs appear like this:
2016-09-01T00:08:16Z|00014|pinctrl|INFO|DHCPOFFER fa:16:3e:25:b0:78 10.0.0.5
2016-09-01T00:08:16Z|00015|pinctrl|INFO|DHCPACK fa:16:3e:25:b0:78 10.0.0.5
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/pinctrl.c | 6 +++
From: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
Add 2 tests for scenarios around lsp-deletion and flow removal
which have escaped current unit tests.
This test depends on the following patch:
"ovn-controller: Back out incremental processing" and passes
after applying it, but
From: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
The following leaks are due to missing ds_destroy in a few
places in build_acl.
5,850 bytes in 50 blocks are definitely lost in loss record 93 of 93
at 0x4C29BFD: malloc (in
/usr/lib64/valgrind/vgpreload_memcheck-amd64-li
From: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
The following leaks are due to missing ds_destroy in a few
places in build_acl.
5,850 bytes in 50 blocks are definitely lost in loss record 93 of 93
at 0x4C29BFD: malloc (in
/usr/lib64/valgrind/vgpreload_memcheck-amd64-li
ome of the data. This fixes the problem.
>
> CC: Chandra S Vejendla <csvej...@us.ibm.com>
> Reported-by: Ramu Ramamurthy <ramu.ramamur...@gmail.com>
> Fixes: 8439c2ebd823 ("ovn: Support for GARP for NAT IPs via localnet")
> Signed-off-by: Ben Pfaff <b...@ovn.org>
The following are among many reported by
"make check-valgrind" on the ovn tests. The
ones in pinctrl.c and binding.c seem real.
2246/valgrind.28645-==28647== 1,496 bytes in 22 blocks are definitely
lost in loss record 175 of 179
2246/valgrind.28645-==28647==at 0x4C29BFD: malloc (in
From: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
This commit fixes the following leak found by check-valgrind in the test:
"send gratuitous arp for nat ips in localnet"
sset gets allocated but not destroyed.
valgrind.14154-==14157== 1,892 bytes in 44 blocks are definitely lost
t;
> - A logical flow which implements the DHCP reponder by sending
>the DHCPv4 reply back to the inport once the 'put_dhcp_opts' action
>is applied.
>
> Signed-off-by: Numan Siddique <nusid...@redhat.com>
> Co-authored-by: Ben Pfaff <b...@ovn.org>
&g
Tested to work end-to-end with openstack.
On Sat, Jul 23, 2016 at 5:49 AM, Numan Siddique wrote:
> OVN implements a native DHCPv4 support which caters to the common
> use case of providing an IP address to a booting instance by
> providing stateless replies to DHCPv4
Tested-By: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
with corresponding openstack patch.
Is there a mechanism to rate-limit the rate of broadcast
dhcp-requests ? This is useful when there are untrusted vifs requesting dhcp.
I sent dhcp-requests from a namespace at a high rate (100
Tested-By: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com> against the
corresponding (rebased) openstack patch:
https://review.openstack.org/#/c/243174/
Hence, Acked-By: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
On Sun, Jul 3, 2016 at 8:54 PM, Numan Siddique <nusid...@redhat.co
I tested and verified this patchset using the corresponding WIP
openstack dhcp patch.
native-dhcp works end-to-end as advertized - I tested with the default
dhcp-options
that are used when VMs boot.
On Mon, Jun 6, 2016 at 10:49 PM, Numan Siddique wrote:
> OVN implements a
Numan, I tested this patchset along with the corresponding WIP ml2
native-dhcp openstack patch.
It looks good to me,
On Tue, May 24, 2016 at 11:04 AM, Numan Siddique wrote:
> OVN implements a native DHCP support which caters to the common
> use case of providing an IP
Tested-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
I tested this patchset to work end-to-end with openstack (with the
corresponding WIP openstack patch).
> +/* Check that the DHCP Message Type (opt 53) is present or not with
> + * valid values - DHCP_MSG_DISCOVER or DHCP
> The reason for not adding the flow in IN_ACL is because the CMS can add
> flows to allow or drop DHCP traffic on a logical port if it wants to. In
> the case of OpenStack networking-ovn, it is adding the below flows for each
> logical port.
>
> table=4( ls_in_acl), priority= 2002,
Darrell, Ben,
Thanks for the fix, The warning was introduced by my patch "send GARP
on localnet"
In the future, I will run my changes through sparse (and possibly
clang) to detect such
problems prior to sharing a patch.
ovn/controller/pinctrl.c:609:39: warning: incorrect type in assignment
Tested-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
I tested v5 of this patchset to work end-to-end with openstack (using
your openstack changes which are also under review). The options tested
include dns-server and classless-static-route.
A question I have is why you program these o
/testsuite.dir/2035/hv/br-int.mgmt:
connected
2016-04-28T18:35:42.460Z|00027|rconn|INFO|unix:/home/ovs/ovs/tests/testsuite.dir/2035/hv/br-int.mgmt:
connected
2016-04-28T18:35:42.460Z|00028|ovsdb_idl|INFO|ovsdb_idl_row_create:
Now, the transaction is complete
Signed-off-by: Ramu Ramamurthy
>
> Is there an equivalent of GARP for IPv6?
>
The "Unsolicited Neighbor Advertisement" seems to be the
equivalent of GARP for IPv6.
http://tools.ietf.org/html/rfc2461#section-7.2.6
The problem motivating this fix was seen with IPv4 though.
___
dev
/networking-ovn/+bug/1545897
Reported-by: Kyle Mestery <mest...@mestery.com>
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/ovn-controller.c | 2 +-
ovn/controller/pinctrl.c| 244 +++-
ovn/controller/pinctrl.h
Move the function extract_lport_addresses to a file
in ovn/lib since that function can be used by ovn-controller also
to parse addresses stored in the mac column of the
port_binding table. Currently that function is used only
in ovn_northd.
Signed-off-by: Ramu Ramamurthy <ramu.rama
The first patch refactors the function extract_lport_addresses
so that it can be used in ovn-northd and in ovn-controller.
The second patch depends on the first patch, and implements the
functionality to send gratuitous arps on localnet.
Changes v1->v2:
Changes per code-review:
* send
Tested-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
Numan, I tested this patch to work on devstack+ovn without the openstack-plugin,
with manual configuration.
Notes:
1) In ovn/utilities/ovn-nbctl.c, usage() Can you add a help string to
ovn-nbctl for the new command
lswitch-get-dhcp-o
On Mon, Apr 18, 2016 at 2:55 PM, Aaron Rosen wrote:
> I like this idea as well.
>
> The one question I have with is is how we should determine which ip address
> to select for the 'distributed' port?
Aaron, Thanks for your review, and feedback.
We can use the dhcp-port
>
> neutron-ovn-metadata would be something we maintain in our OpenStack plugin
> repo (networking-ovn), right?
>
Russell, Thanks for your review and feedback,
neutron-ovn-metadata would be maintained in networking-ovn repo.
> I like this proposal. It suggests adding only the minimal amount of
> This is an interesting proposal. The alternative, which strikes me as a
> terrible idea, would be to somehow implement a TCP/IP stack and HTTP
> server inside OVS.
>
> Does there need to be a single metadata server per hypervisor, or one
> per logical switch per hypervisor?
>
Ben, Thanks for
> +/* dhcp options */
> +shash_init(_opt_symtab);
> +dhcp_opt_expr_symtab_add_field(_opt_symtab, "offerip", 0,
> + DHCP_OPT_TYPE_IP4);
> +dhcp_opt_expr_symtab_add_field(_opt_symtab, "netmask", 1,
> +
On Mon, Apr 11, 2016 at 6:12 PM, Ben Pfaff <b...@ovn.org> wrote:
> On Sun, Apr 03, 2016 at 09:49:03PM -0400, Ramu Ramamurthy wrote:
>> Move the function extract_lport_addresses to a file
>> in ovn/lib since that function can be used by ovn-controller also
>> to parse
On Mon, Apr 11, 2016 at 6:29 PM, Ben Pfaff <b...@ovn.org> wrote:
> On Sun, Apr 03, 2016 at 09:49:04PM -0400, Ramu Ramamurthy wrote:
>> In some usecases such as VM migration or when VMs reuse IP addresses,
>> VMs become unreachable externally because
>> external switches/
On Tue, Apr 12, 2016 at 9:37 AM, Darrell Ball <dlu...@gmail.com> wrote:
> Ramu
>
> High level comment earlier that still applies:
>
> 1) garp reply is still being used rather than garp request
>garp replies are more likely to be dropped or explicitly filtered so gar
27
--state_path=/opt/stack/data/neutron --metadata_port=80
--metadata_proxy_user=1001 --metadata_proxy_group=1001 --debug --verbose
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/binding.c | 73 +--
ovn/controller
On Mon, Apr 4, 2016 at 5:58 AM, Russell Bryant wrote:
> This patch implements one approach to using ovn-controller to implement
> a software l2 gateway between logical and physical networks.
>
> A new logical port type called "gateway" is introduced here. It is very
> close to
> @@ -89,10 +89,11 @@ enum ovn_stage {
> PIPELINE_STAGE(SWITCH, IN, PORT_SEC_L2,0, "ls_in_port_sec_l2") \
> PIPELINE_STAGE(SWITCH, IN, PORT_SEC_IP,1, "ls_in_port_sec_ip") \
> PIPELINE_STAGE(SWITCH, IN, PORT_SEC_ND,2, "ls_in_port_sec_nd") \
> -
/networking-ovn/+bug/1545897
Reported-by: Kyle Mestery <mest...@mestery.com>
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/ovn-controller.c | 2 +-
ovn/controller/pinctrl.c| 217 +++-
ovn/controller/pinctrl.h
Move the function extract_lport_addresses to a file
in ovn/lib since that function can be used by ovn-controller also
to parse addresses stored in the mac column of the
port_binding table. Currently that function is used only
in ovn_northd.
Signed-off-by: Ramu Ramamurthy <ramu.rama
>
> On a separate but related topic, northd programs localnet to not respond to
> regular
> arp requests coming from the network via a localnet port.
>
> How can IP traffic be initiated from a VM outside OVN control or from the
> physical network to a VM inside OVN controlled HV, connected via a
On Thu, Mar 31, 2016 at 12:33 PM, Numan Siddique wrote:
>>
>> +/* Add a new vif to announce garps */
>> +static void
>> +send_garp_add(const struct sbrec_port_binding *binding_rec,
>> + int ofport)
>> +{
>> +int i;
>> +for (i = 0; i < binding_rec->n_mac;
>
> Why did you choose to use reply packets ?
>
> request packets are less likely to be rejected by external networks
> and I think thats discussed in the RFC
>
>
I wanted to use the GARP request, but found a problem.
OVN currently programs flows to respond to gratuitous arp requests from VMs,
add a test to validate garp generation on localnet
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
tests/ovn.at | 61
1 file changed, 61 insertions(+)
diff --git a/tests/ovn.at b/tests/ovn.at
index f2ceba3..7
Generate gratuitous ARP replies when a logical
port on a localnet gets added. Such gratuitous arps
help to update the mac-port bindings and arp-caches
of external switches and routers on the localnet.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/ovn-contro
Move the function extract_lport_addresses under ovn/lib
since that function can be used by ovn-controller also
to parse addresses stored in the mac column of the
port_binding table. Currently that funtion is used only
in ovn_northd.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.
Problem:
When a VM comes up with interface on localnet, external
switches on the broadcast segment have to update their
port->mac mappings and routers have to update their arp caches.
This normally happens when the VM initiates traffic.
on the localnet interface. But in some scenarios, when the
On Wed, Mar 23, 2016 at 4:23 PM, Ben Pfaff <b...@ovn.org> wrote:
> On Tue, Mar 22, 2016 at 04:37:16PM -0700, Ramu Ramamurthy wrote:
>> > This adds a test for this invariant for a specific test case, which is
>> > useful. However, I believe that this should be a unive
> This adds a test for this invariant for a specific test case, which is
> useful. However, I believe that this should be a universal invariant
> for ovn-controller, so that at any time if ovn-controller is restarted
> gracefully after it has converged, it should preserve these flows. Is
> that
Update documentation for the new external-id
in the interface record of ovsdb. The key consists if the
string "ovn-zone-id" concatenated with the logical port id,
and the value is the zone-id.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/ovn-
Add a test to validate that a restart of the ovn-controller
will program zoneids consistently on ports. Flows before
restart are compared against flows after restart to detect
problems with ofports or zone-ids.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
tes
.
Reported-by: Russell Bryant <russ...@ovn.org>
Reported-at: https://bugs.launchpad.net/networking-ovn/+bug/1538696
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/binding.c | 157 ++-
1 file changed, 141 insertions(+),
Changes v5 to v6
* updates per code-review
* handle child-ports and localnet ports
* add a test
* update docs
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 2016-03-15 11:16, Andy Zhou wrote:
On Mon, Mar 14, 2016 at 12:11 PM, Ramu Ramamurthy <
srama...@linux.vnet.ibm.com> wrote:
When a logical switch (localnet) has lots of ports on a hypervisor,
I find that broadcast packets from one of the ports is only forwarded
to
a subset of the
On 2016-03-16 16:35, Andy Zhou wrote:
On Wed, Mar 16, 2016 at 4:00 PM, Ramu Ramamurthy <
srama...@linux.vnet.ibm.com> wrote:
On 2016-03-15 11:16, Andy Zhou wrote:
On Mon, Mar 14, 2016 at 12:11 PM, Ramu Ramamurthy <
srama...@linux.vnet.ibm.com> wrote:
When a logical switch (l
On 2016-03-15 11:16, Andy Zhou wrote:
On Mon, Mar 14, 2016 at 12:11 PM, Ramu Ramamurthy <
srama...@linux.vnet.ibm.com> wrote:
When a logical switch (localnet) has lots of ports on a hypervisor,
I find that broadcast packets from one of the ports is only forwarded
to
a subset of the
When a logical switch (localnet) has lots of ports on a hypervisor,
I find that broadcast packets from one of the ports is only forwarded to
a subset of the other ports, and the kernel module shows the message -
"kernel: openvswitch: ovs-system: deferred action limit reached, drop
recirc action"
> What I would probably do is submit them together: the zone-ids fix + this
> test case that helps test it. If you take out your "xyz" change, will this
> test fail currently?
>
> If i take out the change which masks the zone-ids, this test fails on
master because, zone-ids dont match before and
>
> Does this depend on your zone-ids fix? Do you have another rev of that
> coming?
>
Russell, This test does not depend on the zone-id fix, because it replaces
zone-ids on flows
with a dummy string "xyz". It passes on the master as is.
I will send the next revision of the zone-id fix soon,
Add a test to validate that a restart of the ovn-controller
will program flows in table 0 consistently. Flows before
restart are compared against flows after restart to detect
problems with ofports or zone-ids.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
tes
ion.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/patch.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/ovn/controller/patch.c b/ovn/controller/patch.c
index 753ce3e..cfae613 100644
--- a/ovn/controller/patch.c
+++ b/ovn/controller/patch.c
Thanks for your review!
>
> From this part of the patch, it looks like external-ids:zone-id is
> accepted even if there are duplicates, or if the value is not valid. I
> think that it should reject such cases:
>
> With that in mind, update_local_zone_ids() should also update
>
Reported-At: https://bugs.launchpad.net/networking-ovn/+bug/1539347
Reported-By: na...@cn.ibm.com
Co-Authored-By: na...@cn.ibm.com,rua...@cn.ibm.com
---
openstack-neutron (CMS) allows the configuration of
"static-routes" on a router as shown below.
neutron router-update --route
>
>
> Unfortunately, I thought of another issue that complicates this whole
> approach. A single interface does not necessarily map to a single
> logical port and zone ID. We support sub-ports, initially aimed at
> modelling containers in VMs. That means we need to track N different
> zone IDs
rom that record
Reported-by: Russell Bryant <russ...@ovn.org>
Reported-at: https://bugs.launchpad.net/networking-ovn/+bug/1538696
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
Changes v3 to v4: update as per code-review
* simplify code to update zone-id o
rom that record
Reported-by: Russell Bryant <russ...@ovn.org>
Reported-at: https://bugs.launchpad.net/networking-ovn/+bug/1538696
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
Changes v3 to v4: update as per code-review
* simplify code to update zone-id only when ne
rom that record
Reported-by: Russell Bryant <russ...@ovn.org>
Reported-at: https://bugs.launchpad.net/networking-ovn/+bug/1538696
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
Changes v1 to v2
* add check for null br-int
Changes v2 to v3
* updates per code-review
-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/binding.c | 70
+++-
1 file changed, 63 insertions(+), 7 deletions(-)
diff --git a/ovn/controller/binding.c b/ovn/controller/binding.c
index ce9cccf..28adcef 100644
---
Resend (apologies) as the mailer mangled spaces on the patch.
Signed-off-by: Ramu Ramamurthy <ramu.ramamur...@us.ibm.com>
---
ovn/controller/binding.c | 70
+++-
1 file changed, 63 insertions(+), 7 deletions(-)
diff --git a/ovn/cont
On 2016-02-08 11:12, Russell Bryant wrote:
On 02/08/2016 02:00 PM, Ramu Ramamurthy wrote:
Resend (apologies) as the mailer mangled spaces on the patch.
Can you try posting with the git-send-email utility?
I am posting again from a better behaved mailer
Problem:
-
We are running centos 7.2 (ovs-kernel module from distribution) with OVS
2.3.1.
Large tcp frames (> 30Kbytes gro'd) are coming into a vxlan-tunnel port
and switched by an ovs-bridge onto a veth port
vxlan-tunnel--(gro'd tcp frames sized > 30K)-->OVS-bridge--(frames
On 2016-01-13 16:22, Jesse Gross wrote:
On Wed, Jan 13, 2016 at 3:51 PM, Ramu Ramamurthy
<srama...@linux.vnet.ibm.com> wrote:
Problem:
-
We are running centos 7.2 (ovs-kernel module from distribution) with
OVS
2.3.1.
Large tcp frames (> 30Kbytes gro'd) are coming int
On 2015-08-27 18:26, Jesse Gross wrote:
On Thu, Aug 27, 2015 at 5:46 PM, Ramu Ramamurthy
srama...@linux.vnet.ibm.com wrote:
We found that enabling UDP checksums for vxlan triggers GRO on the
receiver,
and boosts the vxlan performance to 8+ Gbps.
Hence, we want to enable vxlan csum for OVS
posting again, since the earlier emailer mangles email addresses.
Overview:
We are running openstack kilo with neutron OVS using intel 10G
adapters (br kx4 dual-port, 82599es).
neutron OVS is using vxlan tunnels. We are using RHEL 7.1 (which uses
the 3.x kernel).
Openvswitch kernel
79 matches
Mail list logo