lVD\PQ¯d$l
7õJg#È0qm,lâµÎÛ1Wç¸:É2å7¥FZrnü§vTÔÙóÏ¥{8PÔã×&ºÆg¦N¿9_F_¶[MÕ3\!L´]Ó0óÇ,ýÆÜïTÀ'¿'áybT¤kQ,¹ãÎeJ"KACðñøN\Û²ÒdYQè¯ïw
ÛZëç '¡%mô;쥯ÕCDò
x{Ò°ß)OÔÚ¢rç3 RéóÉôSÕFÅ<îb
isåfrLóGƨĬ¤e¨m<Ù»¢ç0ó#"q¶û#H$î×æ·WÖ~ßË~ßm¢·¸Ï_ìuû5gÏ~'_DÖÔ÷H1®lhÖ±äpu¶&
Ž#)ß3ôÌ"$HwÖ!J5
On Mon, Jan 25, 2016 at 03:16:22PM +0200, Alexandru Ardelean wrote:
> On Mon, Jan 25, 2016 at 10:13 AM, Alexandru Ardelean > wrote:
>
> >
> >
> > On Sun, Jan 24, 2016 at 11:24 PM, Russell Bryant wrote:
> >
> >> On 01/24/2016 08:56 AM, Alexandru Ardelean wrote:
> >> > Seems that musl libc's defau
On Sun, Jan 24, 2016 at 03:56:40PM +0200, Alexandru Ardelean wrote:
> Seems that musl libc's default thread stack size is 80k, which
> causes a segfault (stack overflow actually) when trying to add
> a bridge (via the "ovs-vsctl add-br" command).
>
> OpenWRT has been switching to musl libc for a f
Hi Aaron,
I didn't follow all of the list yet - thanks for pointing that other series
out.
That looks even more cleanly integrated, so I agree to wait until that
series is accepted - hopefully before 2.5 is released.
Christian Ehrhardt
Software Engineer, Ubuntu Server
Canonical Ltd
On Mon, Jan 25
Currently tx_qid is equal to pmd->core_id. This leads to unexpected
behavior if pmd-cpu-mask different from '/(0*)(1|3|7)?(f*)/',
e.g. if core_ids are not sequential, or doesn't start from 0, or both.
Example:
starting 2 pmd threads with 1 port, 2 rxqs per port,
pmd-cpu-mask =
Introducing the new model of rx queue management and
fixing unexpected behavior connected with bad distribution
of tx queues.
Complete descriptions are in commit-messages.
Version 4:
* Condition variable returned.
* Style fixes.
Version 3:
* Changed synchronization between
Current rx queue management model is buggy and will not work properly
without additional barriers and other syncronization between PMD
threads and main thread.
Known BUGS of current model:
* While reloading, two PMD threads, one already reloaded and
one not yet reloaded, can poll
On Tue, Jan 26, 2016 at 01:37:05PM +0800, Na Zhu wrote:
> Is there IRC meeting on freenode for ovn, if yes, what is the time?
10:15 am US Pacific time, Thursdays, #openvswitch.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listi
Hi,
Is there IRC meeting on freenode for ovn, if yes, what is the time?
BR
Juno
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Thanks for review.
I'll prepare new version.
Best regards, Ilya Maximets.
On 26.01.2016 03:21, Daniele Di Proietto wrote:
> Hi,
>
> Thanks again for the patch,
>
> As I said in the discussion of v2, I think we should keep the
> condition variable, so that we can return from the datapath
> func
On 26.01.2016 03:21, Daniele Di Proietto wrote:
> Hi,
>
> On 25/01/2016 04:00, "Ilya Maximets" wrote:
>
>> Thank you.
>> Comments inline.
>>
>> On 23.01.2016 04:21, Daniele Di Proietto wrote:
>>> Hi Ilya,
>>>
>>> Thank you very much for the patch.
>>>
>>> I definitely like that the queue assignm
EL0R'S, MASTER'S & DOCT0RATE DEGREES
Add Bache1or's, Master's or Doctorate Degrees to your resume in just a few
weeks and open avenues to promotion and better jobs!
At your Own Pace!
At your Own Schedule!
At your Own Convenience!
No Examination!
No Study!
No Class!
Regardless of your age, sex,
On 25/01/2016 18:20, "Andy Zhou" wrote:
>On Sun, Jan 24, 2016 at 8:32 AM, Ben Pfaff wrote:
>
>> Before this commit, emc_processing() copied a netdev_flow_key if there
>>was
>> no exact-match cache (EMC) hit. This commit eliminates the copy by
>> constructing the netdev_flow_key in the place i
Apologies for the delay.
I applied this to master and branch-2.5 with the following incremental,
to remove a deprecation warning:
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index de7e488..31e56b6 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netdev-dpdk.c
@@ -1480,9 +1480,9 @@ netdev_dpdk_get_
On 1/21/2016 1:46 AM, Traynor, Kevin wrote:
>> -Original Message-
>> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Aaron Conole
>> Sent: Tuesday, January 19, 2016 1:28 PM
>> To: Qiu, Michael
>> Cc: dev@openvswitch.org; Flavio Leitner; Zoltan Kiss
>> Subject: Re: [ovs-dev] [PAT
On Sun, Jan 24, 2016 at 8:32 AM, Ben Pfaff wrote:
> Before this commit, emc_processing() copied a netdev_flow_key if there was
> no exact-match cache (EMC) hit. This commit eliminates the copy by
> constructing the netdev_flow_key in the place it would be copied.
>
> Found by inspection.
>
> Sha
Hi,
Thanks again for the patch,
As I said in the discussion of v2, I think we should keep the
condition variable, so that we can return from the datapath
functions when the port operations are effectively completed.
I also have a couple of minor coding style suggestions (inline).
Other than th
Hi,
On 25/01/2016 04:00, "Ilya Maximets" wrote:
>Thank you.
>Comments inline.
>
>On 23.01.2016 04:21, Daniele Di Proietto wrote:
>> Hi Ilya,
>>
>> Thank you very much for the patch.
>>
>> I definitely like that the queue assignment is performed by the
>> main thread: not only is less bug-prone
Get a Degree in 5 weeks with our program!
~ Our program will let ANYONE with professional experience gain a Degree:
~ Doctorate
~ Bachelors
~ Masters
Just think about it... Follow YOUR Dreams!
Live a wonderful life by earning or upgrading your degree.
This is a good chance to make a right move
On Fri, Jan 22, 2016 at 6:17 PM, Han Zhou wrote:
>
> Regarding the functionality of port-security itself, I am not sure how
would it be supported for ls_out_port_sec. If a dst MAC is not recognised
in ls_in_l2_lkup stage, it is meaningless to have it allowed in
ls_out_port_sec, because the packet
Previously, all ct() actions applied to localnet ports used the default
conntrack zone. We should allocate a ct zone ID for all localnet ports
just like we do for all local VIFs so that none of our connection
tracking interferes with any base system connection tracking in the
default zone.
Signed
v3->v4:
- Added patch #3, which fixes another localnet bug I came across.
I added it to this existing series as the fix depends on changes
in this series.
Russell Bryant (3):
ovn-controller: Move local_datapaths calculation.
ovn: Fix localnet ports on the same chassis.
ovn-controller:
Multiple logical ports on the same chassis that were connected to the
same physical network via localnet ports were not able to send packets
to each other. This was because ovn-controller created a single patch
port between br-int and the physical network access bridge and used it
for all localnet
Before this patch, physical.c build up the set of local datapaths for
its own use. I'd like to use it in another module in a later patch, so
pull it out of physical. It's now populated by the bindings module,
since that seems like a more appropriate place to do it, and it's also
done much earlier
On Mon, Jan 25, 2016 at 11:24:18AM -0800, Jarno Rajahalme wrote:
> > On Jan 25, 2016, at 2:40 AM, Jan Scheurich
> > wrote:
> > 1. OpenFlow bundle depends on OF 1.4. Not many controllers support
> > OF 1.4 and its bundle mechanism yet. Our proposal has no such
> > dependency.
>
> Any new extensio
Russell Bryant wrote on 25/01/2016 10:06:35 PM:
> On 01/16/2016 03:16 AM, Liran Schour wrote:
> > Python idl works now with "monitor_cond" method. Add test
> > for backward compatibility with old "monitor" method.
> >
> > Signed-off-by: Liran Schour
>
> As requested by Andy Zhou, this review
On 01/16/2016 03:16 AM, Liran Schour wrote:
> Python idl works now with "monitor_cond" method. Add test
> for backward compatibility with old "monitor" method.
>
> Signed-off-by: Liran Schour
As requested by Andy Zhou, this review is focused on Python specific
issues, and not necessarily a revie
Current implementation of dpdk_dev_parse_name does not perform a robust
error handling, port names as "dpdkr", "dpdkr1x", "dpdkr 5" are considered
valid.
With this path only positive port numbers in decimal notation are considered
valid.
Signed-off-by: Mauricio Vasquez B
---
v2:
- replace strtol
Current implementation of the ring_client test only supports up to the
dpdkr255 port, this patch extends it to support the full range of possible
dpdkr ports.
Signed-off-by: Mauricio Vasquez B
---
v2:
- replace strtol by strtoul
Thanks to Aaron Conole!
tests/dpdk/ring_client.c | 44
On Mon, Jan 25, 2016 at 3:09 AM, Liran Schour wrote:
> The proposed API for monitor_cond_update that accepts 2 different
> conditions: added and removed is too complex and does not simplify client
> usage.
> Seems like changing monitor_cond_update to receive a single condition will
> simplify th
Thanks, applied to master.
On Mon, Jan 25, 2016 at 11:27:57AM -0800, Jarno Rajahalme wrote:
> Acked-by: Jarno Rajahalme
>
> > On Jan 25, 2016, at 8:40 AM, Ben Pfaff wrote:
> >
> > Signed-off-by: Ben Pfaff
> > ---
> > FAQ.md | 19 +++
> > 1 file changed, 19 insertions(+)
> >
>
Acked-by: Jarno Rajahalme
> On Jan 25, 2016, at 8:40 AM, Ben Pfaff wrote:
>
> Signed-off-by: Ben Pfaff
> ---
> FAQ.md | 19 +++
> 1 file changed, 19 insertions(+)
>
> diff --git a/FAQ.md b/FAQ.md
> index 29b2e19..15056b0 100644
> --- a/FAQ.md
> +++ b/FAQ.md
> @@ -1979,6 +1979,2
On Mon, Jan 25, 2016 at 10:25:15AM -0800, Jarno Rajahalme wrote:
> Thanks for catching this, I’m not sure if I was aware of resubmit to the
> current table feature when I wrote that.
>
> Acked-by: Jarno Rajahalme
Thanks. I applied this to master, branch-2.5, and branch-2.4.
___
OVS already has the notion of “generation ID” on each flow table entry, in the
form of a version range.
Overall, I’d rather see the bundle mechanism extended to cover meters and
groups (i.e., extend the group and meter tables to support the same version
range mechanism). More comments below.
>
On 22 January 2016 at 07:12, Lance Richardson wrote:
> Revert previous workaround for netcat portability issue.
>
> This reverts commit dc55e9465511dee6c12dbf0edb4ce2d9af57cb15.
>
> Signed-off-by: Lance Richardson
Applied to master, thanks!
___
dev mai
On 22 January 2016 at 07:12, Lance Richardson wrote:
> Test cases using netcat ("ICMP related" and "ICMP related with NAT")
> currently fail on systems using the nmap version of nc because this
> version does not support the -q command-line option.
>
> Fix this by detecting which version of netcat
On 25 January 2016 at 10:42, Aaron Conole wrote:
> Mauricio Vasquez B writes:
>
> > Current implementation allows to support only until the dpdkr255 port,
> > this patch extends it to support the full range of possible dpdkr ports.
>
> I think this commit message is not correct. This patch just
Hello Aaron,
On 25 January 2016 at 10:28, Aaron Conole wrote:
> Mauricio Vasquez B writes:
>
> > Current implementation of dpdk_dev_parse_name does not perform a robust
> > error handling, port names as "dpdkr" or "dpdkr1x" are considered valid.
>
> Mauricio, thanks for the patch!
>
> > Signed-
Thanks for catching this, I’m not sure if I was aware of resubmit to the
current table feature when I wrote that.
Acked-by: Jarno Rajahalme
> On Jan 22, 2016, at 3:59 PM, Ben Pfaff wrote:
>
> When recirculation defers actions for processing later, it decides
> based on the actions being saved
On Sun, Jan 10, 2016 at 11:18 PM, Pravin B Shelar wrote:
> diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
> index cd72e62..c746cc2 100644
> --- a/lib/dpif-netdev.c
> +++ b/lib/dpif-netdev.c
> +static inline unsigned long long
> +cycles_counter(void)
> +{
> +#ifdef DPDK_NETDEV
> +return rte
On Mon, Jan 25, 2016 at 11:41:33AM -0500, Russell Bryant wrote:
> On 01/23/2016 08:17 PM, Ben Pfaff wrote:
> > Suggested-by: William Tu
> > Signed-off-by: Ben Pfaff
> > ---
> > lib/ovs-rcu.h | 6 --
> > 1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/lib/ovs-rcu.h b/lib
On 01/23/2016 08:17 PM, Ben Pfaff wrote:
> Suggested-by: William Tu
> Signed-off-by: Ben Pfaff
> ---
> lib/ovs-rcu.h | 6 --
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/lib/ovs-rcu.h b/lib/ovs-rcu.h
> index 5cd4152..7fe7d47 100644
> --- a/lib/ovs-rcu.h
> +++ b/lib/ovs
Signed-off-by: Ben Pfaff
---
FAQ.md | 19 +++
1 file changed, 19 insertions(+)
diff --git a/FAQ.md b/FAQ.md
index 29b2e19..15056b0 100644
--- a/FAQ.md
+++ b/FAQ.md
@@ -1979,6 +1979,25 @@ A: When a switch sends a packet to an OpenFlow
controller using a
buffering that the Ope
Hi Daniele,
I reviewed your patches and they look ok.
I also made some regression tests on performance with an IXIA generator. I
would like
to share these results.
I did the following. I applied the patches and ran a couple of tests without
using the
new 'ct' action, so the ConnTracker feature
Hi Christian,
Christian Ehrhardt writes:
> To enable DPDK it needs a positional parameter to ovs-vswitchd.
> Since everybody is encouranged to use the project provided ovs-ctl
> scripts to start/stop the daemons it should allow an external setting of
> these options via an environment variable.
>
2016-01-25 1:00 GMT+08:00 Ben Pfaff :
> On Sun, Jan 24, 2016 at 11:59:57PM +0800, Huanle Han wrote:
> > I think a more line "mirrors &= ~ctx->mirrors;" is needed at the end of
> the
> > "while" loop.
> > Otherwise, duplicated mirror action may be applied in recursive
> > mirror_packet called by ou
Mauricio Vasquez B writes:
> A ring name length of 10 characters is not enough for dpdkr ports
> starting from dpdkr10, then it is increased to RTE_RING_NAMESIZE
> characters.
>
> Signed-off-by: Mauricio Vasquez B
> ---
> lib/netdev-dpdk.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletio
Mauricio Vasquez B writes:
> Current implementation allows to support only until the dpdkr255 port,
> this patch extends it to support the full range of possible dpdkr ports.
I think this commit message is not correct. This patch just extends the
test.
Also, it's not too big a deal, but you've
Mauricio Vasquez B writes:
> Current implementation of dpdk_dev_parse_name does not perform a robust
> error handling, port names as "dpdkr" or "dpdkr1x" are considered valid.
Mauricio, thanks for the patch!
> Signed-off-by: Mauricio Vasquez B
> ---
> lib/netdev-dpdk.c | 22 ++
This series is about exposing more options required for a regular use of OVS
via environment variables.
I consider the first one a requirement for DPDK. The second patch was created
along the way but as of now has no hard use case other than verbosity control.
Never the less I'd think that gives a
ovs-vswitchd has arguments that are currently not exposed by
ovs-ctl as the project provided wrapper to start/stop the daemons.
This patch adds $OVS_VSWITCHD_OPTS which - if set - will be inserted in the
call to ovs-vswitchd as argument(s).
This avoids adding custom vars for each parameter that is
To enable DPDK it needs a positional parameter to ovs-vswitchd.
Since everybody is encouranged to use the project provided ovs-ctl
scripts to start/stop the daemons it should allow an external setting of
these options via an environment variable.
This patch adds $DPDK_OPTS which will be inserted i
Current testing guidelines do not take account of two distinct
OVS builds:
a) 'standard' OVS
b) OVS with DPDK integration
It is critical that all patches are tested against both builds; if not,
there is a strong possibility that code which adversely affects one
of the builds may be upstreame
Currently tx_qid is equal to pmd->core_id. This leads to unexpected
behavior if pmd-cpu-mask different from '/(0*)(1|3|7)?(f*)/',
e.g. if core_ids are not sequential, or doesn't start from 0, or both.
Example:
starting 2 pmd threads with 1 port, 2 rxqs per port,
pmd-cpu-mask =
Current rx queue management model is buggy and will not work properly
without additional barriers and other syncronization between PMD
threads and main thread.
Known BUGS of current model:
* While reloading, two PMD threads, one already reloaded and
one not yet reloaded, can poll
Introducing the new model of rx queue management and
fixing unexpected behavior connected with bad distribution
of tx queues.
Complete descriptions are in commit-messages.
Version 3:
* Changed synchronization between main and pmd thread to
avoid pausing and resuming the threads.
> The 64 bit cookie can be used for storing the generation-id. OVS supports
> blowing away all rules which
> match a cookie value/mask.
As far as I know only flows have cookies in OpenFlow, so this approach cannot
deal with resync of groups and meters.
__
If CONFIG_NUMA disabled in the system, PMD threads can't
be created:
|ovs_numa|INFO|Discovered 0 NUMA nodes and 0 CPU cores
Signed-off-by: Ilya Maximets
---
lib/ovs-numa.c | 36 ++--
1 file changed, 26 insertions(+), 10 deletions(-)
diff --git a/lib/ovs-numa.c b
On 01/25/2016 02:26 AM, Han Zhou wrote:
> Looks good to me, and tested.
Thanks for the feedback!
--
Russell Bryant
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Mon, Jan 25, 2016 at 10:13 AM, Alexandru Ardelean wrote:
>
>
> On Sun, Jan 24, 2016 at 11:24 PM, Russell Bryant wrote:
>
>> On 01/24/2016 08:56 AM, Alexandru Ardelean wrote:
>> > Seems that musl libc's default thread stack size is 80k, which
>> > causes a segfault (stack overflow actually) wh
Thank you.
Comments inline.
On 23.01.2016 04:21, Daniele Di Proietto wrote:
> Hi Ilya,
>
> Thank you very much for the patch.
>
> I definitely like that the queue assignment is performed by the
> main thread: not only is less bug-prone, but the logic will be more
> easily customizable.
>
> I ab
The proposed API for monitor_cond_update that accepts 2 different
conditions: added and removed is too complex and does not simplify client
usage.
Seems like changing monitor_cond_update to receive a single condition will
simplify the API and the code without complicating the client usage.
I wi
>
> I don't see how the proposed feature is better than what we already have. It
> saves a little memory temporarily, but memory isn't
> usually a big deal in typical OVS environments. Is there some other
> advantage?
A couple of points that come into my mind:
1. OpenFlow bundle depends on O
Dont like the job you work at and you think about changing it for something
better?
You have a great chance to do it right now! Best diplomas in every field of
knowledge are now available for your order.
Hurry up to start the new better life with the competitive salary.
Best universities and co
This message was not delivered due to the following reason(s):
Your message was not 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 t
From: Ofer Ben-Yacov
Currently the IDL does not support passive TCP connection,
i.e. when the OVSDB connects to its manager.
This patch enables IDL to use an already-open session
(the one which was previously used for retrieving the db schema).
In addition, it enables IDL to go back to "listen m
On Sun, Jan 24, 2016 at 11:24 PM, Russell Bryant wrote:
> On 01/24/2016 08:56 AM, Alexandru Ardelean wrote:
> > Seems that musl libc's default thread stack size is 80k, which
> > causes a segfault (stack overflow actually) when trying to add
> > a bridge (via the "ovs-vsctl add-br" command).
> >
67 matches
Mail list logo
