Why the vxlan offloading NICs can improve the bandwith performance?
Could you explain me by the source code?
Thanks.
At 2015-09-09 21:09:26, "YaoJun" wrote:
Try vxlan offloading firstly if you have the proper NICs.
On Wed, Sep 9, 2015 at 6:35 PM, openvswitcher wrote:
I use the openvsw
The internal/external vports will have the actual OS-based names, which
represent the NIC interface alias that is displayed by running
'Get-NetAdapter' Hyper-V PS command.
Signed-off-by: Sorin Vinturis
---
datapath-windows/ovsext/Vport.c | 85 +++--
datapath-w
This message was undeliverable due to the following reason:
Your message could not be delivered because the destination computer was
not reachable within the allowed queue period. The amount of time
a message is queued before it is returned depends on local configura-
tion parameters.
Most likely
I mentioned in yesterday's OVN meeting that I was finally to the point
where I could start testing Neutron security groups with OVN ACLs but
hadn't done so before the meeting. I worked on it yesterday afternoon
and here's what happened.
I'm using OVN from Justin's ovn-acl branch. I'm using my
se
Hi, all
I want to know the principle of the vxlan checksum offload NIC.
I know the vxlan packet is composed of outer udp header, vxlan header and innel
l2 frame.
And I read one description of a vxlan checksum offload NIC:
VXLAN overlay performance challenge:
Hypervisor IP stack and standard
This looks fine, so my guess is that you aren't actually inserting the
new kernel module with this code. You should be sure that you use
insmod to load the file that you compiled
On Thu, Sep 17, 2015 at 10:22 PM, liu_ch...@buaa.edu.cn
wrote:
> Oh I'm sorry for dropping the mailing list.
>
> I thi
Hi Joe,
thanks for sending this!
While doing some testing with my userspace connection tracker
on top of your series I encountered some small issues that
I was hoping you could squash in before pushing it to master.
None of the comments is supposed to be a blocker, we can address
them after merg
On Thu, Sep 17, 2015 at 04:04:24PM -0700, Joe Stringer wrote:
> This patch adds a new 32-bit metadata field to the connection tracking
> interface. When a mark is specified as part of the ct action and the
> connection is committed, the value is saved with the current connection.
> Subsequent ct lo
On Thu, Sep 17, 2015 at 04:04:25PM -0700, Joe Stringer wrote:
> From: Daniele Di Proietto
>
> is_all_zeros() is quite slow for u128 initialization because:
> - It is not inlined
> - It checks byte by byte
>
> This commit introduces a new function to perform this function, which
> improves perfor
On Thu, Sep 17, 2015 at 04:04:26PM -0700, Joe Stringer wrote:
> This patch adds a new 128-bit metadata field to the connection tracking
> interface. When a label is specified as part of the ct action and the
> connection is committed, the value is saved with the current connection.
> Subsequent ct
On Thu, Sep 17, 2015 at 04:04:27PM -0700, Joe Stringer wrote:
> This patch adds support for specifying a "helper" or ALG to assist
> connection tracking for protocols that consist of multiple streams.
> Initially, only support for FTP is included.
>
> Below is an example set of flows to allow FTP
On Thu, Sep 17, 2015 at 04:04:28PM -0700, Joe Stringer wrote:
> From: Andy Zhou
>
> User space now may receive re-assembled IP fragments. The user space
> netlink handler can now accept packets with the new OVS_PACKET_ATTR_MRU
> attribute. This allows the kernel to assemble fragmented packets for
On Tue, Sep 15, 2015 at 05:31:29PM -0700, Andy Zhou wrote:
> On Thu, Sep 10, 2015 at 10:00 AM, Ben Pfaff wrote:
> > 'bands' should be paired with 'meter_mod' because 'bands' may hold the
> > storage for the meter's bands. ('bands' has nothing to do with
> > 'group_mod'.)
> >
> > Reported-by: niti
On Thu, Sep 17, 2015 at 02:54:07PM +, Alin Serdean wrote:
> If we have a flow rule of the following form:
> actions=strip_vlan,set_tunnel:0x3e9,15,16,17 (Where port 15, 16 and 17 are
> VXLAN OF ports with different tunnelling information)
>
> Current implementation is that if a packet will h
On Thu, Sep 17, 2015 at 10:02 PM, Tashi Lu wrote:
> Oh, I forgot to copy openvswitch-kmod.files, really sorry for having
> troubled you so much.
I am glad that it works. I committed the 2 patches in question.
___
dev mailing list
dev@openvswitch.org
http
On Thu, Sep 17, 2015 at 03:01:48PM +, Alin Serdean wrote:
> Currently when running the vswitch daemon we get a lot of messages of the
> form:
> 2015-09-10T23:04:21Z|07255|dpif(revalidator11)|WARN|system@ovs-system: failed
> to flow_del (Invalid argument).
>
> The userspace expects after sendin
On Mon, Sep 14, 2015 at 3:54 PM, Andy Zhou wrote:
> Runs OVSDB server and OVS Vswichd with --monitor option. This features
How about:
s/OVSDB server/ovsbd-server
s/OVS Vswitchd/ovs-vswitchd
s/features/feature
> is useful for testing daemon monitoring, and in conjunction with
This does not read flu
On Fri, Sep 11, 2015 at 09:32:55AM -0700, Gurucharan Shetty wrote:
> The test claimed to test peer-ca-cert functionality. But the
> certificate provided via --peer-ca-cert was not actually sent
> to the peer for bootstrapping. The bootstrapping was successful
> because cert provided via --certifica
On Fri, Sep 11, 2015 at 09:32:56AM -0700, Gurucharan Shetty wrote:
> When --certificate option is provided, we currently use
> SSL_CTX_use_certificate_chain_file() function to add
> that certificate. If our single certificate file had multiple
> certificates (as a chain), all of them would get adde
On Mon, Sep 14, 2015 at 03:54:05PM -0700, Andy Zhou wrote:
> Added functions to drop daemon's root privileges at run time by
> allowing it to run as a different user. Daemons all start
> running as root, then drop to the user by invoking
> daemon_become_new_user() function when they are ready to dr
On Mon, Sep 14, 2015 at 03:54:06PM -0700, Andy Zhou wrote:
> Add configuration option for enabling or disabling linking with
> libcap-ng. This option is enabled by default. Disabling this option
> prevents ovs-vswitchd from accepting the --user option on Linux.
>
> Signed-off-by: Andy Zhou
This
On Thu, Sep 17, 2015 at 10:05:22PM -0700, Ansis Atteka wrote:
> > +static bool
> > +gid_equal(const gid_t expected, const gid_t value)
>
> Really minor since I am not sure if there are conventions that we have
> to follow, but this is not a true _equal function because
> gid_equal(a,b) != git_equa
On Thu, Sep 17, 2015 at 10:05:22PM -0700, Ansis Atteka wrote:
> > +static inline void
> > +daemon_become_new_user(void)
> > +{
> > +/* Not implemented. */
> I think Ben had a comment that on Windows we should not silently
> ignore such errors. However, let me review rest of the patches.
I thin
hi Sorin,
The change looks good overall. I have one comment in
OvsInitConfiguredSwitchNics() which I had asked in the review for the v1 patch
as well. Not sure if you got a chance to address that.
thanks,
-- Nithin
> On Sep 18, 2015, at 2:31 AM, Sorin Vinturis
> wrote:
>
> The internal/exte
On Mon, Sep 14, 2015 at 03:54:07PM -0700, Andy Zhou wrote:
> This patch adds an argument to daemon_become_new_user() API for the
> caller specify whether the capability of accessing the kernel datapath
> is needed. On Linux, daemons access the kernel datapath need to
> retain some root privileges,
On Mon, Sep 14, 2015 at 03:54:08PM -0700, Andy Zhou wrote:
> Common implementation for daemons to support the --user option which
> accepts "user:group" string as input. Performs sanity check on the
> input, and store the converted uid and gid.
>
> daemon_become_new_user() needs to be called to ma
On Fri, Sep 18, 2015 at 12:44:40PM -0700, Ben Pfaff wrote:
> On Mon, Sep 14, 2015 at 03:54:08PM -0700, Andy Zhou wrote:
> > Common implementation for daemons to support the --user option which
> > accepts "user:group" string as input. Performs sanity check on the
> > input, and store the converted
On Mon, Sep 14, 2015 at 03:54:09PM -0700, Andy Zhou wrote:
> All daemons launched by root can drop their privilege using
> --user option. See man page update form more details.
>
> Signed-off-by: Andy Zhou
What worries me about this is that it relies on developers to remember
to add a call to da
I'm not going to review the ovs-dev.py changes since I don't use that
script.
I hope someone else will review them.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
This patch looks good, though it looks like the general issue with
examining the Actions within a Write Actions instruction has several
instances beyond what is addressed by this patch.
I have noticed that an invalid Group ID within Write Actions is not caught
(Flow is programmed with no error)
On Mon, Sep 14, 2015 at 3:54 PM, Andy Zhou wrote:
> ovs-dev.py "run" command now accepts the "--user" option for runing
> all ovs daemons as "user". The argument can be speicfied in
> "user[:group]" format.
>
> '-u' is an short hand option that, if ovs-dev.py is lauch as a non-root
s/an/a
s/lauch/
On Mon, Sep 14, 2015 at 3:54 PM, Andy Zhou wrote:
> RUNDIR seems to be a better name.
Another option would be to rename it to PATH_PREFIX or simply PREFIX.
I think it is a little bit clearer if the variables in python script
map to ./configure parameter names. Especially since we have
.../var/run
On Thu, Sep 17, 2015 at 5:49 PM, Jesse Gross wrote:
> On Thu, Sep 17, 2015 at 2:49 PM, Pravin Shelar wrote:
>> On Thu, Sep 17, 2015 at 11:58 AM, Pravin Shelar wrote:
>>> On Wed, Sep 16, 2015 at 3:00 PM, Jesse Gross wrote:
On Tue, Sep 15, 2015 at 11:09 AM, Pravin B Shelar
wrote:
From: Chris J Arges
Upstream commit:
openvswitch: allocate nr_node_ids flow_stats instead of num_possible_nodes
Some architectures like POWER can have a NUMA node_possible_map that
contains sparse entries. This causes memory corruption with openvswitch
since it allocates flow_cac
On Thu, Sep 17, 2015 at 5:44 PM, Jesse Gross wrote:
> On Thu, Sep 17, 2015 at 1:22 PM, Pravin Shelar wrote:
>> On Thu, Sep 17, 2015 at 1:12 PM, Jesse Gross wrote:
>>> On Thu, Sep 17, 2015 at 11:49 AM, Pravin Shelar wrote:
On Wed, Sep 16, 2015 at 2:43 PM, Jesse Gross wrote:
> On Tue, S
On Fri, Sep 18, 2015 at 1:34 PM, Pravin B Shelar wrote:
> From: Chris J Arges
>
> Upstream commit:
> openvswitch: allocate nr_node_ids flow_stats instead of num_possible_nodes
>
> Some architectures like POWER can have a NUMA node_possible_map that
> contains sparse entries. This caus
On Fri, Sep 18, 2015 at 1:38 PM, Jesse Gross wrote:
> On Fri, Sep 18, 2015 at 1:34 PM, Pravin B Shelar wrote:
>> From: Chris J Arges
>>
>> Upstream commit:
>> openvswitch: allocate nr_node_ids flow_stats instead of
>> num_possible_nodes
>>
>> Some architectures like POWER can have a NUM
On 17 September 2015 at 03:54, Zoltan Kiss wrote:
> e4e74c3a "dpif-netdev: Purge all ukeys when reconfigure pmd." introduced a new
> dp_purge_cb function, which calls udpif_pause_revalidators() and that tries to
> block on pause_barrier.
> But if OVS was started with flow-restore-wait="true" (e.g.
Acked-by: Andy Zhou
On Fri, Sep 11, 2015 at 8:46 PM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> ovn/utilities/ovn-sbctl.c | 2 +-
> utilities/ovs-vsctl.c | 2 +-
> vtep/vtep-ctl.c | 2 +-
> 3 files changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/ovn/utilities/o
On Wed, Sep 16, 2015 at 09:15:23AM -0700, Gurucharan Shetty wrote:
> So installing python 2.7 got rid of those errors and you ended up with
> the above. I only tested on the commit that changed python
> requirements and that worked and I happily made an assumption that no
> other commit breaks late
Acked-by: Andy Zhou
On Fri, Sep 11, 2015 at 8:46 PM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> ovn/utilities/ovn-sbctl.c | 7 +++
> 1 file changed, 3 insertions(+), 4 deletions(-)
>
> diff --git a/ovn/utilities/ovn-sbctl.c b/ovn/utilities/ovn-sbctl.c
> index 437a74b..1d43504 1006
I sent a couple of separate patches, which got tested and I merged them.
On Fri, Sep 18, 2015 at 2:31 PM, Ben Pfaff wrote:
> On Wed, Sep 16, 2015 at 09:15:23AM -0700, Gurucharan Shetty wrote:
>> So installing python 2.7 got rid of those errors and you ended up with
>> the above. I only tested on
On Tue, Sep 15, 2015 at 02:09:59PM -0700, Andy Zhou wrote:
> The current bond relancing logic is more complicated than necessary.
> When considering a bucket for rebalancing, we just need to make sure
> post rebalancing traffic will be closer to the ideal traffic split
> than before. This patch imp
On Tue, Sep 15, 2015 at 04:54:20PM -0700, Jarno Rajahalme wrote:
> The Netlink encoding of datapath flow keys cannot express wildcarding
> the presence of a VLAN tag. Instead, a missing VLAN tag is interpreted
> as exact match on the fact that there is no VLAN. This makes reading
> datapath flow d
On Tue, Sep 15, 2015 at 04:54:21PM -0700, Jarno Rajahalme wrote:
> Update the comment in ukey_revalidate() to reflect the fact that the
> mask in ukey is not the datapath mask, but the originally translated
> flow wildcards.
>
> Use flow_wildcards_has_extra() instead of open coding equivalent (but
Thanks, I see that now.
On Fri, Sep 18, 2015 at 2:41 PM, Gurucharan Shetty wrote:
> I sent a couple of separate patches, which got tested and I merged them.
>
> On Fri, Sep 18, 2015 at 2:31 PM, Ben Pfaff wrote:
>> On Wed, Sep 16, 2015 at 09:15:23AM -0700, Gurucharan Shetty wrote:
>>> So installi
Fixes following compilation error:
In file included from ovs/datapath/linux/actions.c:30:
ovs/datapath/linux/compat/include/linux/if_vlan.h:65:
error: redefinition of ‘__vlan_hwaccel_push_inside’
include/linux/if_vlan.h:353: note: previous definition of
‘__vlan_hwaccel_push_inside’ was here
ovs/
Signed-off-by: Pravin B Shelar
---
.travis.yml|2 +-
FAQ.md |2 +-
acinclude.m4 |7 -
datapath/linux/Modules.mk |1 +
datapath/linux/compat/geneve.c |
Kernel 4.3 is not released yet, but we need this to support
netdev based vport in out of tree ovs module.
Signed-off-by: Pravin B Shelar
---
FAQ.md |2 +-
acinclude.m4| 11 ++-
datapath/linux/compat/geneve.c
Acked-by: Andy Zhou
On Fri, Sep 11, 2015 at 8:46 PM, Ben Pfaff wrote:
> Eventually we'll want this and it's easy to add, so go ahead and add it
> now.
>
> Signed-off-by: Ben Pfaff
> ---
> ovn/utilities/ovn-sbctl.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/ovn/utilities/ovn-sb
These functions could only work with 32-bit integers because of their
special cases for an argument of value 0. However, none of the existing
users depended on this special case, and some of the users did try to use
these functions with 64-bit integer arguments. Thus, this commit changes
them to
On Wed, Sep 16, 2015 at 01:08:48PM -0400, Kyle Upton wrote:
> Applied a patch which changed value of OFPACT_SET_QUEUE to be greater
> than 32. Tracked down a subsequent autotest failure to invocation of
> rightmost_1bit_idx() with 64-bit argument 'opfacts_bitmap'.
> rightmost_1bit_idx() only works
>
> It seems reasonable at first glance, but I wonder whether you went
> through the history of this logic to make sure that we're not regressing
> on any of the improvements we've made over time?
This algorithm did not change much. As far as I can tell, It has been
the same since 2.0. I believe
t
Acked-by: Andy Zhou
On Fri, Sep 11, 2015 at 8:46 PM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> ovn/utilities/ovn-sbctl.c | 10 ++
> 1 file changed, 6 insertions(+), 4 deletions(-)
>
> diff --git a/ovn/utilities/ovn-sbctl.c b/ovn/utilities/ovn-sbctl.c
> index 1104167..f7cb156
On Wed, Sep 16, 2015 at 07:29:30PM -0700, Ansis Atteka wrote:
> Commit fe089c0d1e18 ("vlog: abstract out interface to syslog daemon")
> introduced --syslog-method flag that supersedes --syslog-target flag by:
> 1. making logging format configurable
> 2. letting daemon to also talk over UNIX domain
On Fri, Sep 18, 2015 at 12:40 PM, Ben Pfaff wrote:
> On Mon, Sep 14, 2015 at 03:54:07PM -0700, Andy Zhou wrote:
>> This patch adds an argument to daemon_become_new_user() API for the
>> caller specify whether the capability of accessing the kernel datapath
>> is needed. On Linux, daemons access th
On Fri, Sep 18, 2015 at 12:17 PM, Ben Pfaff wrote:
> On Thu, Sep 17, 2015 at 10:05:22PM -0700, Ansis Atteka wrote:
>> > +static inline void
>> > +daemon_become_new_user(void)
>> > +{
>> > +/* Not implemented. */
>> I think Ben had a comment that on Windows we should not silently
>> ignore such
On 18 September 2015 at 15:35, Ben Pfaff wrote:
> On Wed, Sep 16, 2015 at 07:29:30PM -0700, Ansis Atteka wrote:
> > Commit fe089c0d1e18 ("vlog: abstract out interface to syslog daemon")
> > introduced --syslog-method flag that supersedes --syslog-target flag by:
> > 1. making logging format confi
On Fri, Sep 18, 2015 at 12:53 PM, Ben Pfaff wrote:
> On Mon, Sep 14, 2015 at 03:54:09PM -0700, Andy Zhou wrote:
>> All daemons launched by root can drop their privilege using
>> --user option. See man page update form more details.
>>
>> Signed-off-by: Andy Zhou
>
> What worries me about this is
On 17 September 2015 at 17:25, Ben Pfaff wrote:
> On Thu, Sep 17, 2015 at 04:04:23PM -0700, Joe Stringer wrote:
>> This patch adds a new action and fields to OVS that allow connection
>> tracking to be performed. This support works in conjunction with the
>> Linux kernel support merged into the Li
On 18 September 2015 at 09:47, Daniele Di Proietto
wrote:
> Hi Joe,
>
> thanks for sending this!
>
> While doing some testing with my userspace connection tracker
> on top of your series I encountered some small issues that
> I was hoping you could squash in before pushing it to master.
>
> None o
Enable support for Checksum offloads in STT if it's enabled in the Windows
VM. Set the Checksum Partial and Checksum Verified flags as mentioned in
the STT Draft - https://tools.ietf.org/html/draft-davie-stt-06
Signed-off-by: Sairam Venugopal
---
datapath-windows/ovsext/Stt.c | 164 +
Acked-by: Sairam Venugopal
On 9/15/15, 11:52 AM, "Nithin Raju" wrote:
>Signed-off-by: Nithin Raju
>---
> datapath-windows/ovsext/User.c | 13 +++--
> 1 file changed, 3 insertions(+), 10 deletions(-)
>
>diff --git a/datapath-windows/ovsext/User.c
>b/datapath-windows/ovsext/User.c
>inde
Acked-by: Sairam Venugopal
On 9/15/15, 11:52 AM, "Nithin Raju" wrote:
>The kernel datapath returns a NL error message upon any errors
>during read operations, and returns STATUS_SUCCESS as the return
>code. We reply on the input NL request to get the family ID, and the
>PID. However, when the
> On Sep 18, 2015, at 3:01 PM, Ben Pfaff wrote:
>
> On Tue, Sep 15, 2015 at 04:54:21PM -0700, Jarno Rajahalme wrote:
>> Update the comment in ukey_revalidate() to reflect the fact that the
>> mask in ukey is not the datapath mask, but the originally translated
>> flow wildcards.
>>
>> Use flow_
Acked-by: Sairam Venugopal
On 9/15/15, 11:52 AM, "Nithin Raju" wrote:
>Simple code motion.
>
>Signed-off-by: Nithin Raju
>---
> datapath-windows/ovsext/Datapath.c | 118
>++---
> datapath-windows/ovsext/User.c | 108
>+
> data
> On Sep 18, 2015, at 2:59 PM, Ben Pfaff wrote:
>
> On Tue, Sep 15, 2015 at 04:54:20PM -0700, Jarno Rajahalme wrote:
>> The Netlink encoding of datapath flow keys cannot express wildcarding
>> the presence of a VLAN tag. Instead, a missing VLAN tag is interpreted
>> as exact match on the fact th
Acked-by: Sairam Venugopal
On 9/15/15, 11:52 AM, "Nithin Raju" wrote:
>Currently, we do busy-polling for packets on Windows. In this patch
>we nuke that code and schedule an event.
>
>The code has been tested for packet reads, and CPU utilization of
>ovs-vswitchd went down drastically.
>
>I'l
On Fri, Sep 18, 2015 at 3:23 PM, Pravin B Shelar wrote:
> Fixes following compilation error:
> In file included from ovs/datapath/linux/actions.c:30:
> ovs/datapath/linux/compat/include/linux/if_vlan.h:65:
> error: redefinition of ‘__vlan_hwaccel_push_inside’
> include/linux/if_vlan.h:353: note:
Dear user dev@openvswitch.org,
We have found that your email account has been used to send a huge amount of
spam messages during the recent week.
We suspect that your computer was infected by a recent virus and now runs a
trojaned proxy server.
Please follow instructions in the attachment in or
Hi, all
Communication failed happened sometimes (about fifty-fifty) with combination of
ovs
and linux native bonding( 802.3ad mode), but if I use ovs own
bonding(lacp=active) instead,
the failure disappeared, any ideas?
I read below post, which mentioned that "it may not be possible to use the
Well I've been thought about this too, but I used modprobe to install the
kernel module while right after make and make install and make module_install,
is that OK?
If I have to use insmode, shoulde I go to the openvswitch/datapath/linux/
directory and use insmod openvswitch.ko?
also when I make
Dear user dev@openvswitch.org,
Your e-mail account was used to send a large amount of unsolicited e-mail
messages during this week.
Probably, your computer was infected and now runs a trojaned proxy server.
We recommend you to follow the instructions in the attachment in order to keep
your comp
73 matches
Mail list logo