[vpp-dev] more than 65K entries in a FIB table

[Miklos Tirpak]

Hi,

the FIB table implements two reference counters:
/**
 * per-source number of locks on the table
 */
u16 *ft_locks;
u32 ft_total_locks;

While trying to add lots of GTP-U tunnels, which creates a new interface
for each tunnel in a FIB table, we have reached the u16 limit very fast.
Our requirement would be in the million range.

Would you be against of changing the per-source ft_locks to u32?
It increases the memory footprint, but I can hardly imagine having
hundreds of thousands of FIB tables. Adding that many entries into a
single, or few FIB tables is a valid case though.
If you agree, I can create a pull request.

Thanks,
Miklos
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14925): https://lists.fd.io/g/vpp-dev/message/14925
Mute This Topic: https://lists.fd.io/mt/68831062/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

[vpp-dev] Coverity run FAILED as of 2019-12-19 14:02:06 UTC

[Noreply Jenkins]

Coverity run failed today.

Current number of outstanding issues are 2
Newly detected: 0
Eliminated: 0
More details can be found at  
https://scan.coverity.com/projects/fd-io-vpp/view_defects
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14926): https://lists.fd.io/g/vpp-dev/message/14926
Mute This Topic: https://lists.fd.io/mt/68831420/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] QinQ and dot1ad any

[Jon Loeliger via Lists.Fd.Io]

On Wed, Dec 18, 2019 at 2:38 PM John Lo (loj)  wrote:

> Hi Jon,
>
>
>
> You are right on both counts.  It is the combination of dot1q/ad-any and
> exact-match that we should reject.  It is also correct the check should be
> at lower level to reject the combination for both API and CLI.
>
>
>
> Regards,
>
> John
>

John,

A first-effort patch has been submitted here:
https://gerrit.fd.io/r/c/vpp/+/24077

HTH,
jdl
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14927): https://lists.fd.io/g/vpp-dev/message/14927
Mute This Topic: https://lists.fd.io/mt/68757125/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] more than 65K entries in a FIB table

[Neale Ranns via Lists.Fd.Io]

Hi Miklos,

I would not be against it. Please push a patch.

Thanks,
neale

From:  on behalf of Miklos Tirpak 
Date: Friday 20 December 2019 at 00:41
To: "vpp-dev@lists.fd.io" 
Subject: [vpp-dev] more than 65K entries in a FIB table

Hi,

the FIB table implements two reference counters:
/**
 * per-source number of locks on the table
 */
u16 *ft_locks;
u32 ft_total_locks;

While trying to add lots of GTP-U tunnels, which creates a new interface for 
each tunnel in a FIB table, we have reached the u16 limit very fast. Our 
requirement would be in the million range.

Would you be against of changing the per-source ft_locks to u32?
It increases the memory footprint, but I can hardly imagine having hundreds of 
thousands of FIB tables. Adding that many entries into a single, or few FIB 
tables is a valid case though.
If you agree, I can create a pull request.

Thanks,
Miklos
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14928): https://lists.fd.io/g/vpp-dev/message/14928
Mute This Topic: https://lists.fd.io/mt/68831062/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] merge jobs failing..

[Vratko Polak -X (vrpolak - PANTHEON TECHNOLOGIES at Cisco) via Lists.Fd.Io]

> No its not clear to me why it verified…

Because the verification run happened before [2] was merged
(in other words, 23913 should have been rebased after that).

I think this kind of merge error still happens infrequently enough,
and any "remove +1 from affected open changes" ideas
would be voted down as too much hassle.

Vratko.

[2] https://gerrit.fd.io/r/c/vpp/+/23999

From: vpp-dev@lists.fd.io  On Behalf Of Paul Vinciguerra
Sent: Wednesday, December 18, 2019 1:03 AM
To: Ed Kern (ejk) 
Cc: Benoit Ganne (bganne) ; vpp-dev 
Subject: Re: [vpp-dev] merge jobs failing..

Fixed: https://gerrit.fd.io/r/c/vpp/+/24041

On Tue, Dec 17, 2019 at 7:00 PM Ed Kern via Lists.Fd.Io 
mailto:cisco@lists.fd.io>> wrote:

Hey benoit,

Happened to notice (because I was watching the queue anyway since santa damjan 
did a bunch of merges) that ubuntu test/merges started failing
after
https://gerrit.fd.io/r/c/vpp/+/23913

No its not clear to me why it verified…
it looks like a simple typo where you meant to do
from scapy.contrib.geneve import GENEVE
but did
from scapy.layers.geneve import GENEVE

in test/test_trace_filter.py

anyway..all ubuntu merge jobs are failing after that point so if this ISNT the 
error feel
free to flame me and/or the real root cause.

Ed
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14915): https://lists.fd.io/g/vpp-dev/message/14915
Mute This Topic: https://lists.fd.io/mt/68776336/1594641
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  
[pvi...@vinciconsulting.com]
-=-=-=-=-=-=-=-=-=-=-=-
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14929): https://lists.fd.io/g/vpp-dev/message/14929
Mute This Topic: https://lists.fd.io/mt/68776336/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

[vpp-dev] How can I match all traffics in the vnet/classify module? #classify

[Xiaopeng Bai]

Hello, Everyone!
I'm want to match all traffics by using classify module,  but i didn't find any 
way! Does anyone known how to configure it?
I trying to use hex match with mask: 
, bu 
unfortunately, it crashed.
Crashed backtrace:
(gdb) bt
#0  0x75d2e428 in __GI_raise (sig=sig@entry=6) at 
../sysdeps/unix/sysv/linux/raise.c:54
#1  0x75d3002a in __GI_abort () at abort.c:89
#2  0x00407702 in os_panic () at /root/vpp/src/vpp/vnet/main.c:355
#3  0x7618f201 in os_out_of_memory () at 
/root/vpp/src/vppinfra/unix-misc.c:221
#4  0x7619801f in clib_mem_alloc_aligned_at_offset (size=68719476744, 
align=16, align_offset=8, os_out_of_memory_on_failure=1)
at /root/vpp/src/vppinfra/mem.h:143
#5  0x7619824b in vec_resize_allocate_memory (v=0x0, 
length_increment=4294967296, data_bytes=68719476744, header_bytes=8, 
data_align=16)
at /root/vpp/src/vppinfra/vec.c:59
#6  0x76f52a1a in _vec_resize_inline (v=0x0, 
length_increment=4294967296, data_bytes=68719476736, header_bytes=0, 
data_align=16)
at /root/vpp/src/vppinfra/vec.h:147
#7  0x76f540d0 in vnet_classify_new_table (cm=0x77b66b20 
, mask=0x7fffb68b22c0 "", nbuckets=2, memory_size=2097152,
skip_n_vectors=2, match_n_vectors=0) at 
/root/vpp/src/vnet/classify/vnet_classify.c:139
#8  0x76f562da in vnet_classify_add_del_table (cm=0x77b66b20 
, mask=0x7fffb68b22c0 "", nbuckets=2, memory_size=2097152,
skip=2, match=0, next_table_index=4294967295, miss_next_index=4294967295, 
table_index=0x7fffb689eaa8, current_data_flag=0 '\000', current_data_offset=0,
is_add=1, del_chain=0) at /root/vpp/src/vnet/classify/vnet_classify.c:777
#9  0x76f59189 in classify_table_command_fn (vm=0x769013c0 
, input=0x7fffb689ef00, cmd=0x7fffb65b1ed8)
at /root/vpp/src/vnet/classify/vnet_classify.c:1607
#10 0x7661ced3 in vlib_cli_dispatch_sub_commands (vm=0x769013c0 
, cm=0x769015f0 ,
input=0x7fffb689ef00, parent_command_index=669) at /root/vpp/src/vlib/cli.c:649
#11 0x7661cd58 in vlib_cli_dispatch_sub_commands (vm=0x769013c0 
, cm=0x769015f0 ,
input=0x7fffb689ef00, parent_command_index=0) at /root/vpp/src/vlib/cli.c:609
#12 0x7661d398 in vlib_cli_input (vm=0x769013c0 , 
input=0x7fffb689ef00, function=0x766ba8b4 ,
function_arg=0) at /root/vpp/src/vlib/cli.c:750
#13 0x766c1313 in unix_cli_process_input (cm=0x76901da0 
, cli_file_index=0) at /root/vpp/src/vlib/unix/cli.c:2572
#14 0x766c1fd8 in unix_cli_process (vm=0x769013c0 
, rt=0x7fffb685e000, f=0x0) at 
/root/vpp/src/vlib/unix/cli.c:2688
#15 0x76660f6d in vlib_process_bootstrap (_a=140736235005712) at 
/root/vpp/src/vlib/main.c:1475
#16 0x76117f94 in clib_calljmp () at 
/root/vpp/src/vppinfra/longjmp.S:123
#17 0x7fffb54b66e0 in ?? ()
#18 0x76661075 in vlib_process_startup (vm=0x7fffb68b2170, 
p=0x7fffb4f5f388, f=0x7fffb68b2180) at /root/vpp/src/vlib/main.c:1497
Backtrace stopped: previous frame inner to this frame (corrupt stack?)
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14930): https://lists.fd.io/g/vpp-dev/message/14930
Mute This Topic: https://lists.fd.io/mt/68845182/21656
Mute #classify: https://lists.fd.io/mk?hashtag=classify&subid=1480452
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

[vpp-dev] Does IPFIX-EXPORT can export all traffics on an interface? #vnet #ipfix-export

[Xiaopeng Bai]

Hi, guys.
I want to use the vpp to accomplish a product of network traffic analysis, so i 
need to export all traffics IPFIX to the collector.
But as far sa i know, the IPFIX-EXPORT module depends on the classify module, 
and classify module have nothing to match all traffics on an interface,
What should i do in this sutiation?

Best Regards.
Xiaopeng Bai
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14931): https://lists.fd.io/g/vpp-dev/message/14931
Mute This Topic: https://lists.fd.io/mt/68845272/21656
Mute #vnet: https://lists.fd.io/mk?hashtag=vnet&subid=1480452
Mute #ipfix-export: https://lists.fd.io/mk?hashtag=ipfix-export&subid=1480452
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] Does IPFIX-EXPORT can export all traffics on an interface? #vnet #ipfix-export

[Ole Troan]

Xiaopeng,

You can export without depending on the classifier. Check out the flowprobe 
plugin. 

On rx it supports exporting flows from l2, ip4 and/or ip6 input paths. 

Cheers 
Ole

> On 20 Dec 2019, at 07:15, Xiaopeng Bai  wrote:
> 
> Hi, guys.
> I want to use the vpp to accomplish a product of network traffic analysis, so 
> i need to export all traffics IPFIX to the collector.
> But as far sa i know, the IPFIX-EXPORT module depends on the classify module, 
> and classify module have nothing to match all traffics on an interface, 
> What should i do in this sutiation?
> 
> 
> Best Regards.
> Xiaopeng Bai -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> 
> View/Reply Online (#14931): https://lists.fd.io/g/vpp-dev/message/14931
> Mute This Topic: https://lists.fd.io/mt/68845272/675193
> Mute #vnet: https://lists.fd.io/mk?hashtag=vnet&subid=1480722
> Mute #ipfix-export: https://lists.fd.io/mk?hashtag=ipfix-export&subid=1480722
> Group Owner: vpp-dev+ow...@lists.fd.io
> Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [otr...@employees.org]
> -=-=-=-=-=-=-=-=-=-=-=-
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14932): https://lists.fd.io/g/vpp-dev/message/14932
Mute This Topic: https://lists.fd.io/mt/68845272/21656
Mute #vnet: https://lists.fd.io/mk?hashtag=vnet&subid=1480452
Mute #ipfix-export: https://lists.fd.io/mk?hashtag=ipfix-export&subid=1480452
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] Does IPFIX-EXPORT can export all traffics on an interface? #vnet #ipfix-export

[Xiaopeng Bai]

Hi Ole，

Thank you for your reply!

I will checkout the flowprobe plugin and test it .

Best Regards,
Xiaopeng
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14933): https://lists.fd.io/g/vpp-dev/message/14933
Mute This Topic: https://lists.fd.io/mt/68845272/21656
Mute #vnet: https://lists.fd.io/mk?hashtag=vnet&subid=1480452
Mute #ipfix-export: https://lists.fd.io/mk?hashtag=ipfix-export&subid=1480452
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

[vpp-dev] build contiv vswitch production image - kubernetes & contiv vswitch issue

[Mohamed feroz Abdul majeeth]

Hi All,

1. cd CONTIV_DIR*(v3.3.3)* && make contiv-agent contiv-init.
2. cd VPP_DIR(commit id:a6a98b5460f066e0955b89e6782e55ed3e4f0ede).
3. I applied my sample.so plugin patch (I applied my changes manually) to
vpp code.
4. make build-release ; make pkg-deb.
5. cp $VPP_DIR/build-root/*.deb /scratch/
6. cd /scratch && copy
CONTIV_DIR(v3.3.3)/docker/vpp-vswitch/prod/vswitch/govpp.conf
7. cd /scratch && copy
CONTIV_DIR(v3.3.3)/docker/vpp-vswitch/prod/vswitch/vpp.conf
8. cd /scratch && copy
CONTIV_DIR(v3.3.3)/docker/vpp-vswitch/prod/vswitch/vppctl

In scratch folder below files & packages are available:
Dockerfile
vppctl
vpp.conf
govpp.conf
contiv-agent (v3.3.3)
contiv-init (v3.3.3)
libvppinfra-dev_19.08.1-282~ga6a98b546-dirty_amd64.deb
libvppinfra_19.08.1-282~ga6a98b546-dirty_amd64.deb
vpp-dbg_19.08.1-282~ga6a98b546-dirty_amd64.deb
python3-vpp-api_19.08.1-282~ga6a98b546-dirty_amd64.deb
vpp-dev_19.08.1-282~ga6a98b546-dirty_amd64.deb
vpp_19.08.1-282~ga6a98b546-dirty_amd64.deb
vpp-plugin-core_19.08.1-282~ga6a98b546-dirty_amd64.deb
vpp-api-python_19.08.1-282~ga6a98b546-dirty_amd64.deb
vpp-plugin-dpdk_19.08.1-282~ga6a98b546-dirty_amd64.deb

I build a docker image by
dockerfile(CONTIV_DIR(v3.3.3)/docker/vpp-vswitch/prod/vswitch/Dockerfile)
and inside the docker RUN dpkg -i vpp_*.deb vpp-plugin-core_*.deb
vpp-plugin-dpdk_*.deb libvppinfra_*.deb  and made *contiv-vswitch image*.
I deployed above *contiv-vswitch* in kubernetes, when i launch vpp my
sample.so plugin is not loaded. I run the docker images manually and see
"/usr/lib/x86_64-linux-gnu/vpp_plugins", my sample.so plugin is not
available.
I used above same package in bare metal and run same command ( RUN dpkg -i
vpp_*.deb vpp-plugin-core_*.deb vpp-plugin-dpdk_*.deb libvppinfra_*.deb) my
sample.so plugin is loaded when I launch vpp.

why Its behavior is different in container and bare metal.
please let me know if I missed any steps or any additional things need to
be capture.
Pleas help me to solve this problem.

Thanks,
Feroz
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14934): https://lists.fd.io/g/vpp-dev/message/14934
Mute This Topic: https://lists.fd.io/mt/68845816/21656
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-

Re: [vpp-dev] #vpp How can I make SR-IOV VF work in VPP in host ?

[jacicson1987]

This problem is solved.

Just one command:
ip link set enp5s0f1 allmulticast on

So  that VF can receive multicast packets, such as ipv6 neighbor solicitation.. 
IPv6 ND works.
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#14935): https://lists.fd.io/g/vpp-dev/message/14935
Mute This Topic: https://lists.fd.io/mt/68270891/21656
Mute #vpp: https://lists.fd.io/mk?hashtag=vpp&subid=1480452
Group Owner: vpp-dev+ow...@lists.fd.io
Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub  [arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-