Re: [vpp-dev] Freezing Session Deletion Operation

[Rubina Bianchi]

Dear Andrew

My Trex config is uploaded; I also tested the scenario with your Trex config.
The stability of vpp in your run is strange. When I run this scenario, vpp 
crashes in my DUT machine after about 200 second of running Trex.
In this period I see #del sessions is 0 until session pool becomes full, after 
that session deletion starts. But its rate is lower than the one I see when I 
run vpp on single core.

Could you please check my configs once again for any misconfiguration?
Is vpp or dpdk compatible or incompatible with any specified device?

Thanks,
Sincerely

Sent from Outlook

From: Andrew 👽 Yourtchenko 
Sent: Monday, March 12, 2018 1:50 PM
To: Rubina Bianchi
Cc: vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] Freezing Session Deletion Operation

Dear Rubina,

I've tried the test locally using the data that you sent, here is the
output from my trex after 10 minutes running:

-Per port stats table
  ports |   0 |   1
 
-
   opackets |   312605970 |   312191927
 obytes |100919855857 |174147108346
   ipackets |   311329098 |   277120788
 ibytes |173666531289 | 76492053900
ierrors |   0 |   0
oerrors |   0 |   0
  Tx Bw |   1.17 Gbps |   2.01 Gbps

-Global stats enabled
 Cpu Utilization : 21.2  %  30.0 Gb/core
 Platform_factor : 1.0
 Total-Tx:   3.18 Gbps
 Total-Rx:   2.89 Gbps
 Total-PPS   : 901.93 Kpps
 Total-CPS   :  13.52 Kcps

 Expected-PPS: 901.92 Kpps
 Expected-CPS:  13.53 Kcps
 Expected-BPS:   3.18 Gbps

 Active-flows: 8883  Clients :  255   Socket-util : 0.0553 %
 Open-flows  :  9425526  Servers :65535   Socket : 8883
Socket/Clients :  34.8
 drop-rate   :   0.00  bps
 current time: 702.8 sec
 test duration   : 2897.2 sec

So, in my setup worked I could not see the behavior you describe...

But we have at least one more thing that may be different between our
setups - is the trex config.

Here is what mine looks like:

- version: 2
  interfaces: ['03:00.0', '03:00.1']
  port_limit: 2
  memory:
   dp_flows: 200
  port_info:
   - ip: 1.1.1.1
 default_gw: 1.1.1.2
   - ip: 1.1.1.2
 default_gw: 1.1.1.1


Could you send me your trex config to see if that might be the
difference between our setups, so I could try it locally ?

Thanks!

--a

On 3/12/18, Rubina Bianchi  wrote:
> Hi Dear Andrew
>
> I repeated once again my scenarios with short timeouts and upload all
> configs and outputs for your consideration.
> I am clear about that session cleaner process doesn't work properly and my
> Trex throughput  stuck at 0.
> Please repeat this scenario to verify this (Unfortunately vpp is just stable
> for 200 second and after that vpp will be down).
>
> Thanks,
> Sincerely
>
> Sent from Outlook
> 
> From: Andrew 👽 Yourtchenko 
> Sent: Sunday, March 11, 2018 3:48 PM
> To: Rubina Bianchi
> Cc: vpp-dev@lists.fd.io
> Subject: Re: [vpp-dev] Freezing Session Deletion Operation
>
> Hi Rubina,
>
> I am assuming you are observing this both in single core and multicore
> scenario ?
>
> Based on the outputs, this is what I think might be going on:
>
> I am seeing the total# of sessions is 100, and no TCP transient
> sessions - thus the packets that require a a session are dropped.
>
> What is a bit peculiar, is that the session delete# per-worker are
> non-zero, yet the the delete counters are zero. To me this indicates
> there was a fair bit of transient sessions, which also then got
> recycled by the TCP sessions properly established, before the idle
> timeout has expired.
>
> And at the moment of taking the show command output the connection
> cleaner activity has not yet kicked in - I do not see either any
> session deleted by idle timeout nor its timer restarted. Which makes
> me think that the time interval in which you are testing must be
> relatively short...
>
> So, assuming the time between the start of the traffic and the time
> you have 1m sessions is quite short, this is simply using up all of
> the connection pool, a classic inherent resource management issue with
> any stateful scenario.
>
> You can verify that the sessions delete and start building again if
> you issue "clear acl-plugin sessions".
>
> Also, changing the session timeouts to more aggressive values (say, 10
> seconds), should kick off the aggressive connection cleaning, thus
> should unlock this condition. Of course, shorter idle time means
> potentially useful connections removed.  (the commands are "set
> acl-plugin session timeout  idle ").
>
> *if* neither of  the above does not adequately describe what you are
> seeing, the cleaner node
> may for whatever reason ceases

Re: [vpp-dev] make TEST=test_ip6 test failing on multiple machines, when NOT run as root.

[Ed Kern]

Cant/wont make assumptions or guesses as to why the openstack minions exec as 
root.

The container based builds exec as root because thats the default that happens 
when you run a container in general.

if requested/required I could look into the extra steps it would take to have 
it ‘step down’ to a normal user but not sure
at this time how involved of a process that would be to accomplish.

Ed



On Mar 12, 2018, at 2:49 PM, Klement Sekera -X (ksekera - PANTHEON TECHNOLOGIES 
at Cisco) mailto:ksek...@cisco.com>> wrote:

Why does Jenkins run make test under root? Make test does not need root at 
all...

-Original Message-
From: vpp-dev@lists.fd.io 
[mailto:vpp-dev@lists.fd.io] On Behalf Of Dave
Barach (dbarach)
Sent: Monday, March 12, 2018 8:46 PM
To: vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] make TEST=test_ip6 test failing on multiple machines,
when NOT run as root.

Dave Wallace and I were chatting, and he jogged my memory: our Jenkins
minions run "make test" as root.



TEST=test_ip6 fails 100% of the time when run as a normal user. It passes
when run as root.



Would it be possible to fix this class of problem?



Thanks... Dave



From: Dave Barach (dbarach)
Sent: Monday, March 12, 2018 11:51 AM
To: Klement Sekera -X (ksekera - PANTHEON TECHNOLOGIES at Cisco)
mailto:ksek...@cisco.com>>; Maciek Konstantynowicz (mkonstan)
mailto:mkons...@cisco.com>>; Dave Wallace 
mailto:dwallac...@gmail.com>>; Ed Kern
(ejk) mailto:e...@cisco.com>>
Subject: make TEST=test_ip6 test failing on multiple machines...



Folks,



Master/latest as of commit 8d0f2f0031589513833ab726a7bad5d9e19ff898
seems to fail "make TEST=test_ip6 test" 100% of the time, on a couple of
different machines.



Please investigate.



I need to rework some code in src/vnet/ip/ip6_neighbor.c - DGMS - and
there's no point in starting until the ip6 test reliably passes to begin with...



Thanks... Dave

Re: [vpp-dev] make TEST=test_ip6 test failing on multiple machines, when NOT run as root.

[Klement Sekera]

Why does Jenkins run make test under root? Make test does not need root at 
all...

> -Original Message-
> From: vpp-dev@lists.fd.io [mailto:vpp-dev@lists.fd.io] On Behalf Of Dave
> Barach (dbarach)
> Sent: Monday, March 12, 2018 8:46 PM
> To: vpp-dev@lists.fd.io
> Subject: Re: [vpp-dev] make TEST=test_ip6 test failing on multiple machines,
> when NOT run as root.
> 
> Dave Wallace and I were chatting, and he jogged my memory: our Jenkins
> minions run "make test" as root.
> 
> 
> 
> TEST=test_ip6 fails 100% of the time when run as a normal user. It passes
> when run as root.
> 
> 
> 
> Would it be possible to fix this class of problem?
> 
> 
> 
> Thanks... Dave
> 
> 
> 
> From: Dave Barach (dbarach)
> Sent: Monday, March 12, 2018 11:51 AM
> To: Klement Sekera -X (ksekera - PANTHEON TECHNOLOGIES at Cisco)
> ; Maciek Konstantynowicz (mkonstan)
> ; Dave Wallace ; Ed Kern
> (ejk) 
> Subject: make TEST=test_ip6 test failing on multiple machines...
> 
> 
> 
> Folks,
> 
> 
> 
> Master/latest as of commit 8d0f2f0031589513833ab726a7bad5d9e19ff898
> seems to fail "make TEST=test_ip6 test" 100% of the time, on a couple of
> different machines.
> 
> 
> 
> Please investigate.
> 
> 
> 
> I need to rework some code in src/vnet/ip/ip6_neighbor.c - DGMS - and
> there's no point in starting until the ip6 test reliably passes to begin 
> with...
> 
> 
> 
> Thanks... Dave
> 
> 
> 
> 

-=-=-=-=-=-=-=-=-=-=-=-
Links:

You receive all messages sent to this group.

View/Reply Online (#8511): https://lists.fd.io/g/vpp-dev/message/8511
View All Messages In Topic (2): https://lists.fd.io/g/vpp-dev/topic/14491547
Mute This Topic: https://lists.fd.io/mt/14491547/21656
New Topic: https://lists.fd.io/g/vpp-dev/post

Change Your Subscription: https://lists.fd.io/g/vpp-dev/editsub/21656
Group Home: https://lists.fd.io/g/vpp-dev
Contact Group Owner: vpp-dev+ow...@lists.fd.io
Terms of Service: https://lists.fd.io/static/tos
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] make TEST=test_ip6 test failing on multiple machines, when NOT run as root.

[Dave Barach]

Dave Wallace and I were chatting, and he jogged my memory: our Jenkins
minions run "make test" as root.

 

TEST=test_ip6 fails 100% of the time when run as a normal user. It passes
when run as root.

 

Would it be possible to fix this class of problem?

 

Thanks. Dave

 

From: Dave Barach (dbarach) 
Sent: Monday, March 12, 2018 11:51 AM
To: Klement Sekera -X (ksekera - PANTHEON TECHNOLOGIES at Cisco)
; Maciek Konstantynowicz (mkonstan) ;
Dave Wallace ; Ed Kern (ejk) 
Subject: make TEST=test_ip6 test failing on multiple machines...

 

Folks,

 

Master/latest as of commit 8d0f2f0031589513833ab726a7bad5d9e19ff898 seems to
fail "make TEST=test_ip6 test" 100% of the time, on a couple of different
machines.

 

Please investigate. 

 

I need to rework some code in src/vnet/ip/ip6_neighbor.c - DGMS - and
there's no point in starting until the ip6 test reliably passes to begin
with.  

 

Thanks... Dave

 



smime.p7s
Description: S/MIME cryptographic signature

Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency

[John Lo (loj)]

The functions in the area you are interested may be the unformat functions 
unformat_vnet_sw_interface(), unformat_vnet_hw_interface() in 
src/vnet/interface_format.c used by VPP CLI to get sw_if_index or hw_if_index 
from user specified interface name.-John

-Original Message-
From: vpp-dev@lists.fd.io [mailto:vpp-dev@lists.fd.io] On Behalf Of Sara Gittlin
Sent: Monday, March 12, 2018 11:22 AM
To: vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency

btw - is there (like in linux)  a vpp function helper 
"interface-name-to-if-index"  - which its input is interface-name and output  
is vpp-if-index ?

Thank you
-Sara

On Mon, Mar 12, 2018 at 5:16 PM, Sara Gittlin  wrote:
> Thank you Jon
> This indeed very helpful !
> Cheers
> -Sara
>
> On Mon, Mar 12, 2018 at 5:01 PM, Jon Loeliger  wrote:
>>
>>
>> On Mon, Mar 12, 2018 at 3:57 AM, Sara Gittlin 
>> 
>> wrote:
>>>
>>> Dear Florin,
>>>
>>> When using the command " create tap rx-ring-size 4096 tx-ring-size 4096 "
>>>  VPP create the devices with default names tap0, tap1, ..tapn.
>>>
>>> Is it possible for the user to set the name in the 'create' command line ?
>>>
>>> Thank you in advance
>>> -Sara
>>
>>
>> Hi Sara,
>>
>> Welcome to my battle!
>>
>> Luckily, there is a way to do as you wish!  If you set the "instance"
>> value  during the tap_create API call, or use "id " during the 
>> "create tap ..." CLI command, you will create an interface of the 
>> name "tap".  You may not use a different prefix than "tap", but at 
>> least you can know what the name will be when it is created.
>>
>> HTH,
>> jdl
>>
>> 




-=-=-=-=-=-=-=-=-=-=-=-
Links:

You receive all messages sent to this group.

View/Reply Online (#8509): https://lists.fd.io/g/vpp-dev/message/8509
View All Messages In Topic (14): https://lists.fd.io/g/vpp-dev/topic/14091118
Mute This Topic: https://lists.fd.io/mt/14091118/21656
New Topic: https://lists.fd.io/g/vpp-dev/post

Change Your Subscription: https://lists.fd.io/g/vpp-dev/editsub/21656
Group Home: https://lists.fd.io/g/vpp-dev
Contact Group Owner: vpp-dev+ow...@lists.fd.io
Terms of Service: https://lists.fd.io/static/tos
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] #vpp Switching exercise task - VPP progressive tutorial

[Hamid via Lists.Fd.Io]

Hi Burt,

I don't know much about the modules yet. But thanks for trying it out. There 
were no bridge-domain listed (domain 0 is present in the tutorial) so I had to 
view the reference on how to create one.
The tutorial might be a bit old but its Wiki page has been updated a few times 
recently so I thought I could get the implementation details.

Hamid.

Re: [vpp-dev] #vpp Switching exercise task - VPP progressive tutorial

[Burt Silverman]

You have progressed further than I have, Hamid; I am wondering why
vpp_api_test_plugins are able to be loaded early in the tutorial; I guess
that is from 2016, or I don't know what is wrong with my plugins in the
install tree. I mean, their source code does not have VLIB_PLUGIN_REGISTER
like the non test plugins. Anybody can explain it?

Burt

On Mon, Mar 12, 2018 at 2:40 AM, Hamid via Lists.Fd.Io <
14mseesrasool=seecs.edu...@lists.fd.io> wrote:

> Hi,
>
> I am having some issues with the switching exercise of the VPP
> progressive tutorial
> . I have created
> the veth interface between vpp1 and vpp2 the same way it was created
> between vpp1 and host. I also had to create bridge-domain 1 to get the
> output as shown. The problem I am facing is that I cannot get the host, or
> even vpp2 CLI to ping the loop0 interface 10.10.1.2.
>
> Please assist in the matter.
>
> Regards,
> Hamid
> 
>
>

Re: [vpp-dev] A Question about SPAN Dump Details

[John Lo (loj)]

Hi Jon,

That is a good idea. Please submit a patch.

Regards,
John

From: Jon Loeliger [mailto:j...@netgate.com]
Sent: Monday, March 12, 2018 10:54 AM
To: vpp-dev ; John Lo (loj) 
Subject: A Question about SPAN Dump Details

John, et al,

Could we add the "is_l2" flag into the SPAN details message
so that we don't have to maintain global state about what type
of detail is coming back?

Happy to submit that patch if you are OK with the suggestion.

Thanks,
jdl

Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency

[Sara Gittlin]

btw - is there (like in linux)  a vpp function helper
"interface-name-to-if-index"  - which its input is interface-name and
output  is vpp-if-index ?

Thank you
-Sara

On Mon, Mar 12, 2018 at 5:16 PM, Sara Gittlin  wrote:
> Thank you Jon
> This indeed very helpful !
> Cheers
> -Sara
>
> On Mon, Mar 12, 2018 at 5:01 PM, Jon Loeliger  wrote:
>>
>>
>> On Mon, Mar 12, 2018 at 3:57 AM, Sara Gittlin 
>> wrote:
>>>
>>> Dear Florin,
>>>
>>> When using the command " create tap rx-ring-size 4096 tx-ring-size 4096 "
>>>  VPP create the devices with default names tap0, tap1, ..tapn.
>>>
>>> Is it possible for the user to set the name in the 'create' command line ?
>>>
>>> Thank you in advance
>>> -Sara
>>
>>
>> Hi Sara,
>>
>> Welcome to my battle!
>>
>> Luckily, there is a way to do as you wish!  If you set the "instance"
>> value  during the tap_create API call, or use "id " during the
>> "create tap ..." CLI command, you will create an interface of the
>> name "tap".  You may not use a different prefix than "tap", but
>> at least you can know what the name will be when it is created.
>>
>> HTH,
>> jdl
>>
>> 

-=-=-=-=-=-=-=-=-=-=-=-
Links:

You receive all messages sent to this group.

View/Reply Online (#8505): https://lists.fd.io/g/vpp-dev/message/8505
View All Messages In Topic (13): https://lists.fd.io/g/vpp-dev/topic/14091118
Mute This Topic: https://lists.fd.io/mt/14091118/21656
New Topic: https://lists.fd.io/g/vpp-dev/post

Change Your Subscription: https://lists.fd.io/g/vpp-dev/editsub/21656
Group Home: https://lists.fd.io/g/vpp-dev
Contact Group Owner: vpp-dev+ow...@lists.fd.io
Terms of Service: https://lists.fd.io/static/tos
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency

[Sara Gittlin]

Thank you Jon
This indeed very helpful !
Cheers
-Sara

On Mon, Mar 12, 2018 at 5:01 PM, Jon Loeliger  wrote:
>
>
> On Mon, Mar 12, 2018 at 3:57 AM, Sara Gittlin 
> wrote:
>>
>> Dear Florin,
>>
>> When using the command " create tap rx-ring-size 4096 tx-ring-size 4096 "
>>  VPP create the devices with default names tap0, tap1, ..tapn.
>>
>> Is it possible for the user to set the name in the 'create' command line ?
>>
>> Thank you in advance
>> -Sara
>
>
> Hi Sara,
>
> Welcome to my battle!
>
> Luckily, there is a way to do as you wish!  If you set the "instance"
> value  during the tap_create API call, or use "id " during the
> "create tap ..." CLI command, you will create an interface of the
> name "tap".  You may not use a different prefix than "tap", but
> at least you can know what the name will be when it is created.
>
> HTH,
> jdl
>
> 

-=-=-=-=-=-=-=-=-=-=-=-
Links:

You receive all messages sent to this group.

View/Reply Online (#8504): https://lists.fd.io/g/vpp-dev/message/8504
View All Messages In Topic (12): https://lists.fd.io/g/vpp-dev/topic/14091118
Mute This Topic: https://lists.fd.io/mt/14091118/21656
New Topic: https://lists.fd.io/g/vpp-dev/post

Change Your Subscription: https://lists.fd.io/g/vpp-dev/editsub/21656
Group Home: https://lists.fd.io/g/vpp-dev
Contact Group Owner: vpp-dev+ow...@lists.fd.io
Terms of Service: https://lists.fd.io/static/tos
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency

[Jon Loeliger]

On Mon, Mar 12, 2018 at 3:57 AM, Sara Gittlin 
wrote:

> Dear Florin,
>
> When using the command " create tap rx-ring-size 4096 tx-ring-size 4096 "
>  VPP create the devices with default names tap0, tap1, ..tapn.
>
> Is it possible for the user to set the name in the 'create' command line ?
>
> Thank you in advance
> -Sara


Hi Sara,

Welcome to my battle!

Luckily, there is a way to do as you wish!  If you set the "instance"
value  during the tap_create API call, or use "id " during the
"create tap ..." CLI command, you will create an interface of the
name "tap".  You may not use a different prefix than "tap", but
at least you can know what the name will be when it is created.

HTH,
jdl

[vpp-dev] A Question about SPAN Dump Details

[Jon Loeliger]

John, et al,

Could we add the "is_l2" flag into the SPAN details message
so that we don't have to maintain global state about what type
of detail is coming back?

Happy to submit that patch if you are OK with the suggestion.

Thanks,
jdl

Re: [vpp-dev] Freezing Session Deletion Operation

[Andrew Yourtchenko]

Dear Rubina,

I've tried the test locally using the data that you sent, here is the
output from my trex after 10 minutes running:

-Per port stats table
  ports |   0 |   1
 
-
   opackets |   312605970 |   312191927
 obytes |100919855857 |174147108346
   ipackets |   311329098 |   277120788
 ibytes |173666531289 | 76492053900
ierrors |   0 |   0
oerrors |   0 |   0
  Tx Bw |   1.17 Gbps |   2.01 Gbps

-Global stats enabled
 Cpu Utilization : 21.2  %  30.0 Gb/core
 Platform_factor : 1.0
 Total-Tx:   3.18 Gbps
 Total-Rx:   2.89 Gbps
 Total-PPS   : 901.93 Kpps
 Total-CPS   :  13.52 Kcps

 Expected-PPS: 901.92 Kpps
 Expected-CPS:  13.53 Kcps
 Expected-BPS:   3.18 Gbps

 Active-flows: 8883  Clients :  255   Socket-util : 0.0553 %
 Open-flows  :  9425526  Servers :65535   Socket : 8883
Socket/Clients :  34.8
 drop-rate   :   0.00  bps
 current time: 702.8 sec
 test duration   : 2897.2 sec

So, in my setup worked I could not see the behavior you describe...

But we have at least one more thing that may be different between our
setups - is the trex config.

Here is what mine looks like:

- version: 2
  interfaces: ['03:00.0', '03:00.1']
  port_limit: 2
  memory:
   dp_flows: 200
  port_info:
   - ip: 1.1.1.1
 default_gw: 1.1.1.2
   - ip: 1.1.1.2
 default_gw: 1.1.1.1


Could you send me your trex config to see if that might be the
difference between our setups, so I could try it locally ?

Thanks!

--a

On 3/12/18, Rubina Bianchi  wrote:
> Hi Dear Andrew
>
> I repeated once again my scenarios with short timeouts and upload all
> configs and outputs for your consideration.
> I am clear about that session cleaner process doesn't work properly and my
> Trex throughput  stuck at 0.
> Please repeat this scenario to verify this (Unfortunately vpp is just stable
> for 200 second and after that vpp will be down).
>
> Thanks,
> Sincerely
>
> Sent from Outlook
> 
> From: Andrew 👽 Yourtchenko 
> Sent: Sunday, March 11, 2018 3:48 PM
> To: Rubina Bianchi
> Cc: vpp-dev@lists.fd.io
> Subject: Re: [vpp-dev] Freezing Session Deletion Operation
>
> Hi Rubina,
>
> I am assuming you are observing this both in single core and multicore
> scenario ?
>
> Based on the outputs, this is what I think might be going on:
>
> I am seeing the total# of sessions is 100, and no TCP transient
> sessions - thus the packets that require a a session are dropped.
>
> What is a bit peculiar, is that the session delete# per-worker are
> non-zero, yet the the delete counters are zero. To me this indicates
> there was a fair bit of transient sessions, which also then got
> recycled by the TCP sessions properly established, before the idle
> timeout has expired.
>
> And at the moment of taking the show command output the connection
> cleaner activity has not yet kicked in - I do not see either any
> session deleted by idle timeout nor its timer restarted. Which makes
> me think that the time interval in which you are testing must be
> relatively short...
>
> So, assuming the time between the start of the traffic and the time
> you have 1m sessions is quite short, this is simply using up all of
> the connection pool, a classic inherent resource management issue with
> any stateful scenario.
>
> You can verify that the sessions delete and start building again if
> you issue "clear acl-plugin sessions".
>
> Also, changing the session timeouts to more aggressive values (say, 10
> seconds), should kick off the aggressive connection cleaning, thus
> should unlock this condition. Of course, shorter idle time means
> potentially useful connections removed.  (the commands are "set
> acl-plugin session timeout  idle ").
>
> *if* neither of  the above does not adequately describe what you are
> seeing, the cleaner node
> may for whatever reason ceases to kick in every half a second.
>
> To see the dynamics of conn cleaner node, you can use the debug command
> "set acl-plugin session event-trace 1" before the start of the test.
> This will produce the event trace, which you can view by "show
> event-logger all" - this should give a reasonable idea about what the
> cleaner node is up to.
>
> Please let me know.
>
> --a
>
>
>
>
>
> On 3/11/18, Rubina Bianchi  wrote:
>> Hi,
>>
>> I am testing vpp_18.01.1-124~g302ef0b5 (commit:
>> 696e0da1cde7e2bcda5a781f9ad286672de8dcbf) and vpp_18.04-rc0~322-g30787378
>> (commit: 30787378f49714319e75437b347b7027a949700d) using Trex with sfr
>> scenario in one core and multicore state.
>> After a while I saw session deletion rate decreases and vpp throughput
>> becomes 0 bps.
>> All configuration files and outputs are attached.

Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency

[Sara Gittlin]

Dear Florin,

When using the command " create tap rx-ring-size 4096 tx-ring-size 4096 "
 VPP create the devices with default names tap0, tap1, ..tapn.

Is it possible for the user to set the name in the 'create' command line ?

Thank you in advance
-Sara

On Thu, Mar 8, 2018 at 5:32 PM, Florin Coras  wrote:
> Great! :-)
>
> Florin
>
>
> On Mar 8, 2018, at 1:42 AM, Sara Gittlin  wrote:
>
> Florin
> Thank you - i'm done -  i get very good numbers with tap both for
> throughput and latency
>
> -Sara
>
> On Thu, Mar 8, 2018 at 10:55 AM, Sara Gittlin 
> wrote:
>
> Thank you Florin and Dave
> Florin - I can create tap in vpp and tap in namespace - but how to
> connect them like veth-pair ?
> how traffic transmitted from the namespace tap will be received  on
> the vpp tap ? and viceversa
>
> Dave - when i running iperf3 TCP i can see ~ 1000 pps , when i'm
> running UDP i see nothing (checking why...)
>
> Thank you in advance,
> -Sara
>
> On Wed, Mar 7, 2018 at 11:33 PM, Dave Barach  wrote:
>
> “show run” will probably show a very small vector size.
>
>
>
> If so, look at src/vlib/unix/input.c:linux_epoll_input(…). 10ms is exactly
> the epoll_pwait timeout value.
>
>
>
> D.
>
>
>
> From: vpp-dev@lists.fd.io  On Behalf Of Sara Gittlin
> Sent: Wednesday, March 7, 2018 2:02 PM
> To: vpp-dev@lists.fd.io
> Cc: vpp-dev@lists.fd.io
> Subject: Re: [vpp-dev] VPP As A Router Between Namespaces - 10ms latency
>
>
>
> Thank you Hau
>
> i tested w iperf got similar results. I cannot find iperf2. Anyway ns to ns
> directly without vpp is perfect 50 gbps throughput and 10us latency. Tested
> w iperf3. This is very bothering since we decided to go w vpp instead of ovs
>
> Thanks in advance
>
> -Sara
>
>
>
>
>
>
>
> בתאריך 6 במרץ 2018 20:00,‏ "Hao Fu (haof)"  כתב:
>
> I encountered the similar issue before. Try replacing iperf3 with iperf2.
>
> Hao
>
> On 3/6/18, 8:34 AM, "vpp-dev@lists.fd.io on behalf of Sara Gittlin"
>  wrote:
>
>Also the throughput is very poor - iperf3 TCP ~ 2Mbps
>what is wrong here  ?
>
>On Tue, Mar 6, 2018 at 6:20 PM, Sara Gittlin 
> wrote:
>
> Hi,
> i have 2 namespaces connected with veth-pairs to vpp -  see setup here
>
> [https://wiki.fd.io/view/VPP/Configure_VPP_As_A_Router_Between_Namespaces]
>
>
> i see very big latency ~10ms when i ping between the 2 namespaces
> i expected to see latency in the order of 10's us
> ---
> 4 bytes from 172.16.2.2: icmp_seq=1005 ttl=63 time=11.5 ms
> 64 bytes from 172.16.2.2: icmp_seq=1006 ttl=63 time=9.60 ms
> 64 bytes from 172.16.2.2: icmp_seq=1007 ttl=63 time=7.55 ms
> 64 bytes from 172.16.2.2: icmp_seq=1008 ttl=63 time=5.52 ms
> 64 bytes from 172.16.2.2: icmp_seq=1009 ttl=63 time=9.60 ms
> 64 bytes from 172.16.2.2: icmp_seq=1010 ttl=63 time=17.6 ms
> 64 bytes from 172.16.2.2: icmp_seq=1011 ttl=63 time=15.5 ms
> 64 bytes from 172.16.2.2: icmp_seq=1012 ttl=63 time=13.6 ms
> 64 bytes from 172.16.2.2: icmp_seq=1013 ttl=63 time=11.6 ms
> 64 bytes from 172.16.2.2: icmp_seq=1014 ttl=63 time=9.54 ms
> 64 bytes from 172.16.2.2: icmp_seq=1015 ttl=63 time=7.67 ms
> 64 bytes from 172.16.2.2: icmp_seq=1016 ttl=63 time=5.56 ms
> 64 bytes from 172.16.2.2: icmp_seq=1017 ttl=63 time=3.44 ms
>
> 
>
> Who can assist ?
>
> Thanks in advance
> -Sara
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> 

-=-=-=-=-=-=-=-=-=-=-=-
Links:

You receive all messages sent to this group.

View/Reply Online (#8500): https://lists.fd.io/g/vpp-dev/message/8500
View All Messages In Topic (10): https://lists.fd.io/g/vpp-dev/topic/14091118
Mute This Topic: https://lists.fd.io/mt/14091118/21656
New Topic: https://lists.fd.io/g/vpp-dev/post

Change Your Subscription: https://lists.fd.io/g/vpp-dev/editsub/21656
Group Home: https://lists.fd.io/g/vpp-dev
Contact Group Owner: vpp-dev+ow...@lists.fd.io
Terms of Service: https://lists.fd.io/static/tos
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] Freezing Session Deletion Operation

[Rubina Bianchi]

Hi Dear Andrew

I repeated once again my scenarios with short timeouts and upload all configs 
and outputs for your consideration.
I am clear about that session cleaner process doesn't work properly and my Trex 
throughput  stuck at 0.
Please repeat this scenario to verify this (Unfortunately vpp is just stable 
for 200 second and after that vpp will be down).

Thanks,
Sincerely

Sent from Outlook

From: Andrew 👽 Yourtchenko 
Sent: Sunday, March 11, 2018 3:48 PM
To: Rubina Bianchi
Cc: vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] Freezing Session Deletion Operation

Hi Rubina,

I am assuming you are observing this both in single core and multicore
scenario ?

Based on the outputs, this is what I think might be going on:

I am seeing the total# of sessions is 100, and no TCP transient
sessions - thus the packets that require a a session are dropped.

What is a bit peculiar, is that the session delete# per-worker are
non-zero, yet the the delete counters are zero. To me this indicates
there was a fair bit of transient sessions, which also then got
recycled by the TCP sessions properly established, before the idle
timeout has expired.

And at the moment of taking the show command output the connection
cleaner activity has not yet kicked in - I do not see either any
session deleted by idle timeout nor its timer restarted. Which makes
me think that the time interval in which you are testing must be
relatively short...

So, assuming the time between the start of the traffic and the time
you have 1m sessions is quite short, this is simply using up all of
the connection pool, a classic inherent resource management issue with
any stateful scenario.

You can verify that the sessions delete and start building again if
you issue "clear acl-plugin sessions".

Also, changing the session timeouts to more aggressive values (say, 10
seconds), should kick off the aggressive connection cleaning, thus
should unlock this condition. Of course, shorter idle time means
potentially useful connections removed.  (the commands are "set
acl-plugin session timeout  idle ").

*if* neither of  the above does not adequately describe what you are
seeing, the cleaner node
may for whatever reason ceases to kick in every half a second.

To see the dynamics of conn cleaner node, you can use the debug command
"set acl-plugin session event-trace 1" before the start of the test.
This will produce the event trace, which you can view by "show
event-logger all" - this should give a reasonable idea about what the
cleaner node is up to.

Please let me know.

--a





On 3/11/18, Rubina Bianchi  wrote:
> Hi,
>
> I am testing vpp_18.01.1-124~g302ef0b5 (commit:
> 696e0da1cde7e2bcda5a781f9ad286672de8dcbf) and vpp_18.04-rc0~322-g30787378
> (commit: 30787378f49714319e75437b347b7027a949700d) using Trex with sfr
> scenario in one core and multicore state.
> After a while I saw session deletion rate decreases and vpp throughput
> becomes 0 bps.
> All configuration files and outputs are attached.
>
> Thanks,
> Sincerely
>
> Sent from Outlook
>


acls
Description: acls


init.conf
Description: init.conf


show sessions
Description: show sessions


startup.conf
Description: startup.conf


trex
Description: trex