Hi,
My question is more on clarity. It was enabled and working fine. But, some
times I see this behavior.
I need answer on what shall be the output when we enable mpls at interface and
why it shows mpls-not-enabled but it works fine.
when do we see: none configured
when do we see: mpls-not-e
Hi Vpp Expert,
My fdio ver is 21.06. I have added mpls table add 0 and then trying to enable
interface for mpls.
When I enabled: (set interface mpls VirtualFuncEthernet0/7/0.1600 enable)
the output of " show int features VirtualFuncEthernet0/7/0.1600" is as below:
...
mpls-input:
mpls-not-e
Hi Neale,
Thank you so much. I will try to explore today and get back to you.
mean while, Is it that FIB walk is going through loop due to wrong fib objects
or may be we are not destroying the object once processed?
This issue happens only with mpls i.e inner and outer VRFs.
With Regards
Sastry
Hi Neale,
Thank you so much for extending the help.
I am not sure how do I run/configure unit test. I can try If you can provide me
steps to configure VRFs using VL APIs in unit test framework.
We can have a call to discuss?
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all me
Hi Rajith,
Thank you for the help. Even I suspect some issue with mpls forward route.
Could you pls check below?
Labelled route:
vpp# show mpls fib 16
MPLS, fib_index:0 locks:[API:5, ]
16:eos/21 fib:0 index:403 locks:2
API refs:1 entry-flags:uRPF-exempt, src-flags:added,contributing,active,
p
Hi Rajith,
One more imp point here is, when you say " mpls php route without eos" , this
looks like into mpls fib table 0 and my issue is on inner context or inner data
packet.
here Eth interface is more on outer context and IP is inner
context (VRF). I guess we see issue in VRF.
Could you pl
Hi Rajith,
Thank you for the clues but, I am not sure how a route can resolve this issue?
I do not have any data traffic when we see this issue. Just a bgp is up with
vpnv4/vpnv6 and configured follower bfd.
Due to BFD events up/down, its calling adj_bfd_notify which goes through VRF
table i.e
Hi VPP Experts,
Neale try to guide us on this mpls features but, I am not sure if he is around
or may be busy. We are stuck on mpls along with BFD protection.
This is very consistent looping and due to which all the watchdog timers are
getting expired and no clue if its fixed in later releases o
Hi Neale,
The same issue is report little while ago:
vpp-dev@lists.fd.io | Infinite loop in fib_walk_sync (
https://lists.fd.io/g/vpp-dev/topic/83599549#19589 )
Could you please check if its fixed or worked around?. I see that its loop of
vrfs i.e outter and inner context associated and its int
Hi Neale,
Could you please help me on this. I am using fdio 21.06. Some one reported the
same issue and there is no clarity further on this.
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#21045): https://lists.fd.io/g/vpp-dev/m
Hi Vpp Expert,
Could you please help me on the below vpp hang issue while running below
configuration:
Developed App using VPP APIs for the below commands:
set int mpls VirtualFuncEthernet0/8/0.1800 enable
ip table add 2
ip route add table 2 1.1.1.1/32 via 107.243.21.116
VirtualFuncEthernet0/
Hi Naele,
Its resolved with " mp->mr_route.mr_paths[0].sw_if_index = ~0;" .
Thank you.
With regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#20867): https://lists.fd.io/g/vpp-dev/message/20867
Mute This Topic: https://lists.fd.io/m
Hi Neale,
Thank you for your reply. I was offline for sometime and could not track it.
We are using VL APIs to program into vpp and I see some error while programming
mpls route into default mpls table:
Params we pass are as below:
mp->_vl_msg_id = htons (VL_API_MPLS_ROUTE_ADD_DEL);
mp->context
On Mon, Dec 13, 2021 at 08:39 AM, Sastry Sista wrote:
>
> Hi Neale,
> I have used only these 2 below and its not crashing but "MPLS DROP DPO".
>
>
>
> vppctl ip route add table 2 1.1.1.1/32 via 107.243.21.116
> VirtualFuncEthernet0/9/0.1800 out-labels 18
>
[Edited Message Follows]
Hi Neale,
I have used only these 2 below and its not crashing but "MPLS DROP DPO".
vppctl ip route add table 2 1.1.1.1/32 via 107.243.21.116
VirtualFuncEthernet0/9/0.1800 out-labels 18
vppctl mpls local-label 17 1.1.1.2/32 via ip4-lookup-in-table 2
Here, I have created
Hi Neale,
I have used only these 2 below and its crashing.
vppctl ip route add table 2 1.1.1.1/32 via 107.243.21.116
VirtualFuncEthernet0/9/0.1800 out-labels 18
Vppctl mpls local_label 17 1.1.1.2/32 table-id 2
Here, I have created loop18 at table 2 and added 1.1.1.2/24.
Do you say that loop18
Hi Neale,
I tried all combinations of bind ( vppctl mpls local-label 17 1.1.1.2/32). Vpp
is crashing once we receive mpls ping req.
I am using fdio 20.05.
Please confirm if its an open bug:
crashed with below:
=
vppctl ip route add table 2 1.1.1.1/32 via 107.243.21.116
Virtual
Hi Neale,
So, you suggest bind over other way.
Then, I need to do following for bind of a prefix?
vppctl ip route add table 2 1.1.1.1/32 via 107.243.21.116
VirtualFuncEthernet0/9/0.1800 out-labels 18
vppctl mpls local-label 17 eos via ip4-lookup-in-table 2
vppctl mpls local-label 17 1.1.1.2/3
Hi Neale,
Thank you for detailed explanation. Its working fine by providing nh and
interface.
Just a small Q:
without bind API, label to prefix, I could manage to make this working.
I just wanted to know what is the real value of bind API?
Is it that bind will just help us to get label and pre
Hi Neale,
Thank you for the help. We are trying to exercise vpp mpls vpn and we are
preferring label per prefix only. Anyway I am fine even with per VRF label but
adding more local labels which seems fine while adding.
I am trying to evaluate VRF with non-default ip fib tables.
While I tried d
Hi Neale,
Because of this label bind to prefix crash issue, I can think of using label
per VRF and use pop rule as below:
I will add all local labels to same VRF
vppctl mpls local-label 17 eos via ip4-lookup-in-table 1
vppctl mpls local-label 18 eos via ip4-lookup-in-table 1
vppctl mpls local-la
Hi Neale,
I am not able to test bind CLI along with pop rules. I am seeing vpp crash for
the below commands. Using 20.05 fdio.
vppctl set int ip addr VirtualFuncEthernet0/9/0.1800 107.243.21.115/24
vppctl set int state VirtualFuncEthernet0/9/0.1800 up
vppctl mpls table add 0
vppctl set int mpls
Hi Neale,
ping is fine with per vrf label schema.
But, Issue is with label per prefix. I see 20.05 vpp is crashing as below:
Thread 5 "vpp_wk_2" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7f41551fc640 (LWP 101)]
0x7f479f974db4 in ip4_fib_mtrie_lookup_step_one (dst_add
Thank you Neale. We may need to develop something like show mpls bind for our
purpose.
My test setup ping is not working. Error is : "MPLS DROP DPO" because of fib
next node is drop-dpo. COuld you please help me what is worng?
I am trying to try mpls ping between 1.1.1.1/24 to 2.2.2.2/24.
vpp
Hi Neale,
Then, mpls table ID we give for mpls commands always represents ip table id?
for example: mpls local-label 100 del 10.10.20.0/24 table 3 ==> here table is
represents ip fib id?
Also, what is the use of non zero mpls table ID at all? could you please
provide one use case?
Second, if
Hi Neale,
Thank you for the quick reply. I gone through that link and have few queries.
On MPLS VPNs:
Q1. If we do not allow interfaces creation inside VRF of mpls tables, then what
is the use of mpls table 1 etc?
Q2. Always prefixes are associated with IP FIBs as per my understanding, so,
how
Hi,
I am trying l3vpn features and following the below steps:
1. mpls table add 0
2. set interface mpls VirtualFuncEthernet0/7/0.1549 enable
3. mpls table add 1
4. ip route add 2.2.2.2/24 via 10.164.4.75 VirtualFuncEthernet0/7/0.1549
mpls-lookup-in-table 1 out-labels 75
I found 2.2.2.2 in ip fi
Hi Florin,
Sorry for confusing. My point is, whatever we send through VCL APIs, those APIs
execution at VPP inside session layer is at main thread? I assume its executed
at main thread.
Is there anyway we can run the session layer execution for VCL APIs at worker
thread?
I know memif, af_packe
Hi Florin,
Is it that VCL is going to be executed in the context of main thread as its
being connected to VPP main.
Do you see any issue with thread barrier if I send huge data? I know it is
going to impact.
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to th
Hi Neale and Alex,
Thank you for the inputs and Yes, its due to entries in neighbor table.
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#19843): https://lists.fd.io/g/vpp-dev/message/19843
Mute This Topic: https://lists.fd.io/
Hi,
Show ip fib gives may be giving interface and we do search based on interface
on ip neighbor table?
When we have 2 entries in ip neighbor table, how does it work. This is infact a
bug which got fixed with gratuitous ARP.
Could you please let me know the code where exactly looking neighbor t
Hi,
Could anybody help me how does VPP works when we have 2 entries in "show ip
neighbor" table for the same dest addr with different sw_if_index.
Here below is the example: Dest Addr is 107.243.21.113.
vpp# show ip neighbor
Time IP Flags Ethernet
Hi Florin,
Thank you for the pointers. Anyway I did same already as its straightforward.
Hope to no other changes required.
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#19532): https://lists.fd.io/g/vpp-dev/message/19532
Mute
Hi Florin,
Thank you so much and its really great and timely inputs you provided and no
dictionary words can really express my sincere appreciation. Emotionally can be
expressed. Truly amazing. Wish you the good luck.
I have reached final stage to take the decision on the solution. Now that I a
Hi Florin,
Thank you for the bit more clarity. I have few question further to the items:
1. In a single worker case, What is the difference between session_handle from
vppcom_session_create and vppcom_epoll_create.
2. For every TCP session in a worker, we will have 2 different handlers i.e
vp
Hi Florin,
Thank you for the help. I guess its due to INVALID_RMP_IP which is 17. Will
check why its going as for IP. We are filling it fine.
So, My next step would be to implement event handler part of my linux epoll.
I see you suggested me below point in the last mail:
FLORIN : " 2. grab
[Edited Message Follows]
Hi Florin,
Here is show app:
vpp# show app
Index Name Namespace
0 tls default
1 app[shm] default
vpp#
I am running BLOCKED MODE for now just to debug my issue and I see that its hit
at vcl_session_connected_han
[Edited Message Follows]
Hi Florin,
Here is show app:
vpp# show app
Index Name Namespace
0 tls default
1 app[shm] default
vpp#
I am running BLOCKED MODE for now just to debug my issue and I see that its hit
at vcl_session_connected_han
Hi Florin,
Here is show app:
vpp# show app
Index Name Namespace
0 tls default
1 app[shm] default
vpp#
I am running BLOCKED MODE for now just to debug my issue and I see that its hit
at vcl_session_connected_handler with retval as -17 i.
Hi Florin,
app clnts are created successfully as below. Only vppcom_session_connect
returns error within client. Because I am not even hitting vl_api_clnt_process
at server i.e vpp_main.
Could please let me know which is early function that hits in latest version
when vcl client sends connect?
Hi Florin,
Thank you. Yes. I need to upgrade. Or I am thinking of customizing in fdio
20.05 release.
Small question, vcl_test_client is working fine and have session created in my
environment.
But, When I added code client is added to vpp_main like mq and shm.
Issue is, session is returned wit
Hi Florin,
I was on vacation for 2 days and could not reply you.
My requirement is, we will get dynamically vrf from my application and each
session may land in different vrf. As per code, we have option of defining vrf
inside session struct and thats fine right? Why do you think, vcl is not
su
Hi Florin,
Thank you.
Let me be very specific and focused topic:
I have just single worker but having multiple TCP sessions. So, I will poll on
vppcom_mq_epoll_fd() for all notifications. So, notifications are for me to do
data read/write also? If there is a data on shm for any session, we will
Hi Florin,
Thank you so much for all your help. We are very happy to use VPP because of
support you all are giving us.
I am sorry if I am missing anything. If you look at 21.01 version also, We do
not have support for both source address/vrf. We just have those variables
inside session but not
Hi,
lcl_ip and lcl_port looks source/local endpoint details but, I do not see if
we are really copying it to session data:
Please confirm if this source address is supported in 20.05 fdio version?
int
vppcom_session_connect (uint32_t session_handle, vppcom_endpt_t * server_ep)
{
sess
[Edited Message Follows]
Hi,
Could you please confirm if we have support from VCL to control source
sw_if_index. I am using 20.05 fdio version.
I saw the code and found we do not have sw_if_index as below:
session_connect_msg_ Struct Reference
---
Data Fields
---
u32 (
https:
Hi,
Could you please confirm if we have support from VCL to control source
sw_if_index. I am using 20.05 fdio version.
I saw the code and found we do not have sw_if_index as below:
session_connect_msg_ Struct Reference
---
Data Fields
---
u32 (
https://docs.fd.io/vpp/20.05/de/
Hi Florin,
Gone through your old mails on eventfd of mq and given me decent inputs.
vppcom_mq_epoll_fd() is returning mq_epfd which I can use for my app dispatch
linux epoll. But, My concern is, Can I use this mq_epfd for all my TCP sessions
i.e single mq for multiple TCP sessions? I have single
[Edited Message Follows]
Hi Florin,
I found the root cause of that crash. Its due to session token. If I have
session{ xxx...}, VPP crashes as we are always expecting a space session token
and "{ " . May be this problem is common for all tokens.
Regarding my TCP session, I have session is up a
Hi Florin,
I found the root cause of that crash. Its due to session token. If I have
session{ xxx...}, I crash as we are always expecting a space session token and
"{ " . May be this problem is common for all tokens.
Regarding my TCP session, I have session is up and could send data over it. I
Hi Florin,
Once I have session info as below, I am not seeing vpp crash.
session {
evt_qs_memfd_seg
}
I am good now and not seeing unsupported application config message also. If I
have any other param or empty params, vpp crashes. I think, we need to address
this in later vpp versions. COuld
Hi Florin,
Both of these are not hit. So, is it config object crash?
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#19348): https://lists.fd.io/g/vpp-dev/message/19348
Mute This Topic: https://lists.fd.io/mt/82575076/21656
Mute
Hi Florin,
I am running in gdb only.
last few lines in gdb:
fdio/src/fdio.2005/build-root/ install-vpp_debug-native/vpp/ bin/vpp[2240]:
dpdk: EAL init args: -c 6 -n 4 --in-memory --file-prefix vpp -w :00:06.0 -w
:00:09.0 --master-lcore 1
[New Thread 0x7ffbf21ff700 (LWP 2244)]
[New Threa
Hi Florin,
I am not seeing even backtrace. bt is just empty now. Also I have just added
session { } with empty params and seeing same exit.
No clues at vpp. is there any params that I need to add along with session so
that it does not exit?
With Regards
Sastry
-=-=-=-=-=-=-=-=-=-=-=-
Links: Yo
Hi Florin,
I tried to add session layer at vpp as below:
session {
session-baseva 0x20
evt_qs_memfd_seg
event-queue-length 10
}
vpp is still crashing and not able to run vpp. Also, is session layer important
for VCL application to run?
I am still seeing "vl_api_app_attach_reply_t_ha
Sastry Sista via lists.fd.io wrote:
> Hi Florin,
> Thank you for the reply.
>
> I am trying to develop similar kind of envoy app using VCL. So, I need
> both socket option for binary API and shared client for rx/tx for TCP data.
> So, I need use-mq-eventfd
Hi Florin,
Thank you for the reply.
I am trying to develop similar kind of envoy app using VCL. So, I need both
socket option for binary API and shared client for rx/tx for TCP data.
So, I need use-mq-eventfd at vcl.conf right?. ANyway I tried removing it
but still same issu
Hi,
I am trying to use vcl_test_client and using below vcl config:
While trying to run seeing the below error:
vppcom_connect_to_vpp:502: vcl<1876:0>: app (vcl_test_client) is connected to
VPP!
vppcom_app_create:1203: vcl<1876:0>: sending session enable
vppcom_app_create:1211: vcl<1876:0>: sendi
Hi Chris,
Thank you for the reply. When vppctl has all the logic to take VAT kind of
commands and get output back except some dynamic events etc.
We wanted to develop VAT kind of application but would like to use socket lib
as transport. VAT is primarily is using shared memeory and VAPI lib also
Hi Ole,
So VAT APIs has only shared memory has transport?
Socket fd APIs like producer_event_fd etc are there but, I am not sure if they
are any use.
Please confirm if there is any other way to use VAT API?
ALso vppctl commands between vpp_main and vppctl is using Unix socket?
With Regards
Sas
Hi,
I am using VAPI c library to use socket APIs to use VAT APIs.
I see that vapi_connect and vapi_send etc socket APIs. I am wondering that
these vapi APIs again using vl_client_connect and svm library calls.
So, how do we say VAPI connect/send using socket fd while sending?
I see still shared
Hi,
We see some issues with shared memory interface for VAT kind of application
that we developed. Issue is we need socket fd so that we can track in our
application.
Now, I see socket interface over VAT is best for our needs.
I see some comment that not all APIs work over socket interface?
Pl
62 matches
Mail list logo
