On 21 July 2017 at 11:16, Joe Stringer wrote:
> On 21 July 2017 at 07:52, Darrell Ball wrote:
>>
>>
>> -Original Message-
>> From: Ilya Maximets
>> Date: Friday, July 21, 2017 at 5:24 AM
>> To: Darrell Ball , Joe
On 21 July 2017 at 06:35, Eric Garver wrote:
> On Fri, Jul 21, 2017 at 04:34:07AM -0700, Tonghao Zhang wrote:
>> The version of ip-route may not support udp6zerocsum for
>> vxlan6 or geneve6. If we run the kernel check, there may
>> be always error message. Before running the test
On 20 July 2017 at 13:02, Darrell Ball wrote:
> Three of the SNAT tests allow for wget retries, which occasionally
> happen. However, these tests did not allow for SNAT address
> variability for the retries, which is now tolerated.
>
> Signed-off-by: Darrell Ball
From: Greg Rose
Upstream commit:
commit 52427fa0631269c62885dc48e0c32e2ad6e17f8c
Author: Daniel Axtens
Date: Mon Jul 3 21:46:43 2017 +1000
openvswitch: fix mis-ordered comment lines for ovs_skb_cb
I was trying to wrap my head around
From: Tonghao Zhang
Upstream commit:
commit 9cc9a5cb176ccb4f2cda5ac34da5a659926f125f
Author: Tonghao Zhang
Date: Thu Jun 29 17:27:44 2017 -0700
datapath: Avoid using stack larger than 1024.
When compiling OvS-master on
Upstream commit:
commit 3fcece12bc1b6dcdf0986f2cd9e8f63b1f9b6aa0
Author: Jakub Kicinski
Date: Fri Jun 23 22:11:58 2017 +0200
net: store port/representator id in metadata_dst
Switches and modern SR-IOV enabled NICs may multiplex traffic from Port
From: Greg Rose
Upstream commit:
commit dc5321d79697db1b610c25fa4fad1aec7533ea3e
Author: Matthias Schiffer
Date: Mon Jun 19 10:03:56 2017 +0200
vxlan: get rid of redundant vxlan_dev.flags
There is no good reason to
From: Greg Rose
Upstream commit cf124db566e6 ("net: Fix inconsistent teardown and
release of private netdev state.") removed the destructor member
of the net_device structure and replaced it with a boolean flag
indicating that the net device resource needs freeing. Use
Upstream commit:
commit de77b966ce8adcb4c58d50e2f087320d5479812a
Author: Johannes Berg
Date: Fri Jun 16 14:29:19 2017 +0200
networking: convert many more places to skb_put_zero()
There were many places that my previous spatch didn't find,
as
From: Greg Rose
Upstream commit:
commit cf124db566e6b036b8bcbe8decbed740bdfac8c6
Author: David S. Miller
Date: Mon May 8 12:52:56 2017 -0400
net: Fix inconsistent teardown and release of private netdev state.
Network devices can
Upstream commit:
commit 7529390d08f07fbf9b0174c5a87600b5caa1a8e8
Author: Davide Caratti
Date: Thu May 18 15:44:42 2017 +0200
openvswitch: more accurate checksumming in queue_userspace_packet()
if skb carries an SCTP packet and ip_summed is
From: Greg Rose
Upstream commit:
commit d91fc59cd77c719f33eda65c194ad8f95a055190
Author: Liping Zhang
Date: Sun May 7 22:01:55 2017 +0800
netfilter: introduce nf_conntrack_helper_put helper function
And convert module_put
This series takes Greg's recent proposed series, rolls in a few missed
upstream commits, and orders them the same as the upstream net-next tree.
Greg, would you mind taking a look at patches #1, #2, #4 and #7? Note that for
some of the patches it wasn't possible to introduce a function in the
Currently the open flow 'clone' action is always translated into
datapath clone. While this is valid translation, the datapath
'clone' action is more expensive and has more restrictions than
not using them.
This patch optimizing the open flow 'clone' translation. Whenever
the open flow actions
Create a new function that hides the details of netlink encoding
for the translated clone action.
Signed-off-by: Andy Zhou
---
ofproto/ofproto-dpif-xlate.c | 76 +++-
tests/ofproto-dpif.at| 7
2 files changed, 39
Use dpif/set-dp-features command instead.
Signed-off-by: Andy Zhou
---
ofproto/ofproto-dpif.c | 34 --
tests/ofproto-macros.at | 5 ++---
2 files changed, 2 insertions(+), 37 deletions(-)
diff --git a/ofproto/ofproto-dpif.c
Use 'dpif/set-dp-features' command instead.
Signed-off-by: Andy Zhou
---
ofproto/ofproto-dpif.c | 21 -
tests/system-traffic.at | 8 ++--
2 files changed, 2 insertions(+), 27 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
When changing support fields, it may be unsafe to set support level
beyond what datapath can support.
This patch introduce the notion of boot time support and
runtime support fields. Boot time support are set only
once during ofproto start up phase, and not changed during
runtime. The runtime
Use 'dpif/set-dp-features' command instead.
Signed-off-by: Andy Zhou
---
ofproto/ofproto-dpif-xlate.c | 11 ---
ofproto/ofproto-dpif-xlate.h | 1 -
ofproto/ofproto-dpif.c | 24
tests/ofproto-dpif.at| 4 +---
4 files changed, 1
dpif support fields contain various datapath capabilities detected
by ofproto at start up time. Usually those fields are read-only,
not intended to be changed at runtime.
However in practice, when writing tests or running experiments, it
becomes necessary to set those fields to emulate different
I respined the patches based on review feedback at:
http://mail.openvswitch.org/pipermail/ovs-dev/2017-June/333502.html
Patch 1-5: Implements a generic command that can defeature and restore
the datapath support level. They are not directly related
to the main objective
On Wed, Jul 19, 2017 at 7:51 AM, Ilya Maximets wrote:
> This allows to compose packets with different real lenghts from
> odp flows i.e. memory will be allocated for requested packet
> size and all required headers like ip->tot_len filled correctly.
>
> Will be used in
On 2 July 2017 at 18:58, Tonghao Zhang wrote:
> Signed-off-by: Tonghao Zhang
Thanks, I'll apply this soon.
___
dev mailing list
d...@openvswitch.org
On 2 July 2017 at 18:57, Tonghao Zhang wrote:
> Upstream commit:
> commit 9cc9a5cb176ccb4f2cda5ac34da5a659926f125f
> Author: Tonghao Zhang
> Date: Thu Jun 29 17:27:44 2017 -0700
>
> datapath: Avoid using stack larger than
En línea y en Vivo / Para todo su Equipo con una sola Conexión
Técnicas Infalibles para Incrementar sus Ventas
04 de Agosto - Online en Vivo - 10:00 a 13:00hrs
Si desea incrementar su rentabilidad, ésta es la oportunidad de equipar a su
fuerza de ventas con un probado y efectivo
Add dev mailing list. It got dropped by accident.
-- Forwarded message --
From: Andy Zhou
Date: Fri, Jul 21, 2017 at 2:14 PM
Subject: Re: [PATCH] bond: Unify hash functions in hash action and entry lookup.
To: Ilya Maximets
As it turns
ssl_new_stream() takes ownership of the passed-in 'name' parameter.
In error scenarios, the name is leaked. I was able to trigger this
leak by attempting to connect to an ovsdb over SSL and specifying
non-existent certificate, private key, and CA cert files.
This patch fixes the problem by
On 21 July 2017 at 11:47, Greg Rose wrote:
> Right, that's a cleaner solution. Thanks for pointing it out. I
> think there are a few others like that in the series then.
OK, I'll keep an eye out for those.
___
dev mailing list
Right, that's a cleaner solution. Thanks for pointing it out. I
think there are a few others like that in the series then.
- Greg
On Fri, Jul 21, 2017 at 11:42 AM, Joe Stringer wrote:
> On 19 July 2017 at 13:57, Greg Rose wrote:
>> Upstream commit:
>>
Yep, missed that. That sounds like the right thing to do.
Thanks,
On Fri, Jul 21, 2017 at 11:41 AM, Joe Stringer wrote:
> On 19 July 2017 at 13:57, Greg Rose wrote:
>> Upstream commit:
>> commit 9cc9a5cb176ccb4f2cda5ac34da5a659926f125f
>> Author:
Sure thing. Thanks Joe!
On Fri, Jul 21, 2017 at 11:41 AM, Joe Stringer wrote:
> On 19 July 2017 at 13:57, Greg Rose wrote:
>> Allow datapath kernel modules to be configured and built for kernels up
>> to 4.12.
>>
>> Adds a new define for the kernel
On 19 July 2017 at 13:57, Greg Rose wrote:
> Upstream commit:
> commit d91fc59cd77c719f33eda65c194ad8f95a055190
> Author: Liping Zhang
> Date: Sun May 7 22:01:55 2017 +0800
>
> netfilter: introduce nf_conntrack_helper_put helper
On 19 July 2017 at 13:57, Greg Rose wrote:
> Upstream commit:
> commit 9cc9a5cb176ccb4f2cda5ac34da5a659926f125f
> Author: Tonghao Zhang
> Date: Thu Jun 29 17:27:44 2017 -0700
>
> datapath: Avoid using stack larger than 1024.
>
>
On 19 July 2017 at 13:57, Greg Rose wrote:
> Allow datapath kernel modules to be configured and built for kernels up
> to 4.12.
>
> Adds a new define for the kernel compatibility layer to indicate whether
> upstream commit cf124db566e6 ("net: Fix inconsistent teardown and
Signed-off-by: Joe Stringer
---
AUTHORS.rst | 1 +
1 file changed, 1 insertion(+)
diff --git a/AUTHORS.rst b/AUTHORS.rst
index 10c84c185262..1370a0dc82c5 100644
--- a/AUTHORS.rst
+++ b/AUTHORS.rst
@@ -316,6 +316,7 @@ Valient Gough vgo...@pobox.com
Venkata Anil
On 6 July 2017 at 13:57, Ben Pfaff wrote:
> From: wangzhike
>
> Ovs and kernel stack would add frag_queue to same netns_frags list.
> As result, ovs and kernel may access the fraq_queue without correct
> lock. Also the struct ipq may be different on kernel(older
On 21 July 2017 at 07:52, Darrell Ball wrote:
>
>
> -Original Message-
> From: Ilya Maximets
> Date: Friday, July 21, 2017 at 5:24 AM
> To: Darrell Ball , Joe Stringer
> Cc: ovs dev , Ben
From: Jan Scheurich
This commit adds support for the OpenFlow actions generic encap
and decap (as specified in ONF EXT-382) to the OVS control plane.
CLI syntax for encap action with properties:
encap(hdr=)
encap(hdr=,
prop(class=,type=,val=),
From: Zoltan Balogh
- Checking decap() prerequisits.
- Encap/decap VLAN tagged Ethernet frames.
- Send L3 packet over patch port.
- Output L2/L3 packet to ports with different packet_type properties.
Signed-off-by: Zoltan Balogh
From: Zoltan Balogh
This series is a continuation of other patch series initiated by Jan Scheurich
before. These were already applied to the master branch:
- userspace: Support for L3 tunneling
https://mail.openvswitch.org/pipermail/ovs-dev/2017-June/87.html
When the interfaces list is retrieved through getiffaddrs(), there
might be elements with iface_name set to NULL.
This patch checks ifa_name to be not NULL before comparing it to the
actual device name in the loop that calculates how many interfaces
exist with that same name.
Also, this patch
-Original Message-
From: Ilya Maximets
Date: Friday, July 21, 2017 at 5:24 AM
To: Darrell Ball , Joe Stringer
Cc: ovs dev , Ben Pfaff
Subject: Re: [ovs-dev] [patch_v1 1/2] System Tests: Allow
Dear Customer,
We can not deliver your parcel arrived at July 19.
Download postal receipt attached to e-mail!
With gratitude,
,
UPS Delivery Manager.
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Fri, Jul 21, 2017 at 04:34:07AM -0700, Tonghao Zhang wrote:
> The version of ip-route may not support udp6zerocsum for
> vxlan6 or geneve6. If we run the kernel check, there may
> be always error message. Before running the test units,
> we check the ip command.
>
> Signed-off-by: Tonghao
'lookup_bond_entry' currently uses 'flow_hash_symmetric_l4' while
OVS_ACTION_ATTR_HASH uses 'flow_hash_5tuple'. This may lead to
inconsistency in slave choosing for the new flows. In general,
there is no point to unify hash functions, because it's not
required for correct work, but it's logically
Russell Bryant writes:
> On Thu, Jul 20, 2017 at 3:41 AM, Numan Siddique wrote:
>> On Wed, Jul 19, 2017 at 7:00 PM, Aaron Conole wrote:
>>
>>> Timothy Redaelli writes:
>>>
>>> > The policy is to use %files to
When a new outermost MPLS label is added to 'flow' the 'flow''s
Ethernet type is changed to 'mpls_eth_type'. After the new label is
set, the 'flow''s MPLS stack is updated, and the L3/4 fields are
cleared to mark them invalid. This results in loosing the values of
the 'nw_tos' and the 'nw_ttl'
There are many reasons for reloading of pmd threads:
* reconfiguration of one of the ports.
* Adjusting of static_tx_qid.
* Adding new tx/rx ports.
In many cases EMC is still useful after reload and uninit
will only lead to unnecessary upcalls/classifier lookups.
Such
Reconfiguration of HW NICs may lead to packet drops.
In current model all physical ports will be reconfigured each
time number of PMD threads changed. Since we not stopping
threads on pmd-cpu-mask changes, this patch will help to further
decrease port's downtime by setting the maximum possible
Currently, change of 'pmd-cpu-mask' is very heavy operation.
It requires destroying of all the PMD threads and creating
them back. After that, all the threads will sleep until
ports' redistribution finished.
This patch adds ability to not stop the datapath while
adjusting number/placement of PMD
Version 4:
* Trivial rebase of patch #2.
Version 3:
* Added comment about 'static_txq_id's adjustment.
* Added additional parentheses around 'dynamic_txqs'
comparison operand because of warning from GCC 6.3.1.
Version 2:
* Dropped patch [1/4] as already
On 21.07.2017 05:40, Darrell Ball wrote:
> The discussion about the ‘Area’ prefix has come up again, even after Ben had
> commented about it
> and after I had pointed folks to the submitting-patches.rst, which allows
> flexibility in choosing an
> ‘Area’ prefix by the patch submitter.
>
> In
The version of ip-route may not support udp6zerocsum for
vxlan6 or geneve6. If we run the kernel check, there may
be always error message. Before running the test units,
we check the ip command.
Signed-off-by: Tonghao Zhang
---
tests/system-common-macros.at | 10
Rxqs consumed processing cycles are used to improve the balance
of how rxqs are assigned to pmds. Currently some reconfiguration
is needed to perform a reassignment.
Add an ovs-appctl command to perform a new assignment in order
to balance based on the latest rxq processing cycle information.
Up to his point rxqs are sorted by processing cycles they
consumed and assigned to pmds in a round robin manner.
Ian pointed out that on wrap around the most loaded pmd will be
the next one to be assigned an additional rxq and that it would be
better to reverse the pmd order when wraparound
Previously rxqs were assigned to pmds by round robin in
port/queue order.
Now that we have the processing cycles used for existing rxqs,
use that information to try and produced a better balanced
distribution of rxqs across pmds. i.e. given multiple pmds, the
rxqs which have consumed the largest
Count the cycles used for processing an rxq during the pmd
optimization interval. As this is an in flight counter and
pmds run independently, also store the total cycles used
during the last full interval.
Signed-off-by: Kevin Traynor
---
lib/dpif-netdev.c | 30
Add two counters to dp_netdev_rxq which will be used for storing the
processing cycles of an rxq. Processing cycles will be stored in reference
to a defined interval. One counter is used for storing cycles during the
current in progress interval, while the other is used to store the cycles
of the
Soon we will want to store processing cycle counts in the dp_netdev_rxq,
so use that as a basis for the polled_queue that pmd_thread_main uses.
Signed-off-by: Kevin Traynor
---
lib/dpif-netdev.c | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git
So far the interval was only used for dpcls optimization.
Soon, we will use it for storing rxq cycles so make the
names more generic. Also, set the interval regardless
of whether dpcls optimization has occurred, as the
optimization interval will need to be consistent across
pmds.
Signed-off-by:
For the DPDK datapath, by default rxqs are assigned to available pmds
in round robin order with no weight or priority.
It can happen that some very busy queues are handled by one pmd which
does not have enough cycles to prevent packets being dropped on them.
While at the same time another pmd
>> Keepalive feature is aimed at achieving Fastpath Service Assurance in
>> OVS-DPDK deployments. It adds support for monitoring the packet
>> processing cores(PMD thread cores) by dispatching heartbeats at
>> regular intervals. Incase of heartbeat misses additional health checks
>> are enabled on
Keepalive feature is aimed at achieving Fastpath Service Assurance
in OVS-DPDK deployments. It adds support for monitoring the packet
processing cores(PMD thread cores) by dispatching heartbeats at regular
intervals. Incase of heartbeat misses additional health checks are
enabled on the PMD thread
This commit adds the support to check the PMD cycle stats. If the cycles
aren't changing for a duration of time this can be flagged as possible
PMD stall.
Signed-off-by: Bhanuprakash Bodireddy
---
lib/dpif-netdev.c | 17 +
lib/dpif-netdev.h | 7
The keepalive thread sends heartbeats to PMD thread and when PMD fails to
respond to successive heartbeats the PMD is potentially stalled. The PMD
state transition is as below:
ALIVE -> MISSING -> DEAD -> GONE
This commit enables PMD healthchecks when PMD doesn't respond to
heartbeats. This is
This commit allows vswitchd thread to update the OVSDB with the
status of all registered PMD threads. The status can be monitored
using ovsdb-client and the sample output is below.
$ ovsdb-client monitor Open_vSwitch Open_vSwitch keepalive
rowaction keepalive
This commit adds the support to check the packet statistics on the port
polled by PMD thread. If the packets aren't processed due to PMD thread
stall/deadlock the statistics wont update and this can be used by
monitoring framework to confirm PMD failure.
This mechanism has limitation with MQ
This commit adds support to query keepalive statistics. Datapath health
status can be retrieved as follows:
$ ovs-appctl keepalive/pmd-health-show
Keepalive status
keepalive status : Enabled
keepalive interval: 1000 ms
PMD threads : 8
PMDCORESTATE
This commit adds the initial support in to performing PMD health checks.
The ports handled by the PMD threads are checked for the link status and
the same is updated in to keepalive info structure.
Signed-off-by: Bhanuprakash Bodireddy
---
lib/dpif-netdev.c |
This commit registers the packet processing PMD cores to keepalive
framework. Only PMDs that have rxqs mapped will be registered and
actively monitored by KA framework.
This commit spawns a keepalive thread that will dispatch heartbeats to
PMD cores. The pmd threads respond to heartbeats by
This commit enables additional datapath health checks. The checks
are enabled only on a PMD heartbeat failure. On missing three successive
heartbeats additional health checks needs to be performed on respective
PMD thread to confirm the failure.
The datapath health is monitored periodically from
This commit adds heartbeat mechanism support for DPDK datapath. Heartbeats
are sent to registered PMD threads at predefined intervals (as set in ovsdb
with 'keepalive-interval').
The heartbeats are only enabled when there is atleast one port added to
the bridge and with active PMD thread polling
This commit implements APIs to retrieve the PMD thread status and return
the status in the below format for each PMD thread.
Format: pmdid="status,core id,last_seen_timestamp"
eg: pmd62="ALIVE,2,9220698256784207"
pmd63="GONE,3,9220698256786231"
The status is periodically
This commit introduces the initial keepalive support by adding
'keepalive' module and also helper and initialization functions
that will be invoked by later commits.
This commit adds new ovsdb column "keepalive" that shows the status
of the datapath threads. This is implemented for DPDK datapath
Implement helper function to retrieve the process status. This commit
also enables the fields relating to process name, state and core the
process was last scheduled. The APIs will be used by keepalive monitoring
framework in future commits.
Signed-off-by: Bhanuprakash Bodireddy
This commit fixes the following "sparse" warning:
lib/process.c:439:16: error: use of assignment suppression and length
modifier together in gnu_scanf format [-Werror=format=].
This fix doesn't need any other changes as the fields aren't used for now.
Signed-off-by: Bhanuprakash Bodireddy
Introduce helper functions in 'dpdk' module that are needed for
DPDK keepalive functionality. Also add dummy functions in 'dpdk-stub' module
that are needed when DPDK datapath is not available.
Signed-off-by: Bhanuprakash Bodireddy
---
lib/dpdk-stub.c | 24
Keepalive feature is aimed at achieving Fastpath Service Assurance
in OVS-DPDK deployments. It adds support for monitoring the packet
processing cores(PMD thread cores) by dispatching heartbeats at regular
intervals. Incase of heartbeat misses additional health checks are
enabled on the PMD thread
I did some testing; display looks nice to me
Other comments inline
> 2017-07-21T06:38:33.215Z|00053|unixctl|DBG|received request
> dpctl/ct-bkts["netdev@ovs-netdev","gt=0"], id=0
> 2017-07-21T06:38:33.215Z|00054|dpctl|INFO|set_names=0 verbosity=0 names=0
>
Sorry for the delay, I missed this one which makes perfect sense.
Acked-by: Miguel Angel Ajo
On Sun, Jul 16, 2017 at 9:56 PM, Russell Bryant wrote:
> The OVN gateway HA design document is very useful in its current form.
> It describes a range of options
80 matches
Mail list logo