Hi Steven,
It turns out that moving to VPP v1710 resolves the issue and ping latency. I am
still using my own kernel images with VPP v1710.
Thanks for you help. I may have more queries related to vhost-user.
Nitin
________________________________
From: vpp-dev-boun...@lists.fd.io <vpp-dev-boun...@lists.fd.io> on behalf of
Saxena, Nitin <nitin.sax...@cavium.com>
Sent: Monday, October 2, 2017 9:51 PM
To: Steven Luong (sluong)
Cc: Narayana, Prasad Athreya; vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] [vhost-user][armv8][v17.07.01] VM ping not working
properly
Hi Steven,
Thanks for your help. I can understand your restrictions. Btw I am trying to
use ubuntu cloud image with VPP. Will let you know the result.
Thanks,
Nitin
________________________________
From: Steven Luong (sluong) <slu...@cisco.com>
Sent: Saturday, September 30, 2017 7:28 AM
To: Saxena, Nitin
Cc: Narayana, Prasad Athreya; vpp-dev@lists.fd.io
Subject: Re: [vpp-dev] [vhost-user][armv8][v17.07.01] VM ping not working
properly
Nitin,
I am using the bios and the cloud image which I downloaded directly from the
Ubuntu web site
-bios QEMU_EFI.fd
-device virtio-blk-device,drive=image \
-drive if=none,id=image,file=1-xenial-server-cloudimg-arm64-uefi1.img \
You are using the kernel that you built from your internal repos and the img
file, ubuntu-aarch64-ext4.img, which you created from ubuntu xenial 16.04.
-kernel 2-cagit1-4.12.9-virtio-4K-Image
-drive file=2-ubuntu-aarch64-ext4.img
I only have ThunderX, but your problem may be specific to ThunderX2 as you
described. You can try stable/1710 to see if it helps because that is what I
use. I am afraid that I don’t have any more tricks in my sleeve to offer useful
help. Sorry about that.
Steven
From: "Saxena, Nitin" <nitin.sax...@cavium.com>
Date: Thursday, September 28, 2017 at 6:56 PM
To: "Steven Luong (sluong)" <slu...@cisco.com>
Cc: "Narayana, Prasad Athreya" <prasadathreya.naray...@cavium.com>,
"vpp-dev@lists.fd.io" <vpp-dev@lists.fd.io>
Subject: Re: [vpp-dev] [vhost-user][armv8][v17.07.01] VM ping not working
properly
Hi steven,
I am working on ThunderX2 which is different from ThunderX in many ways.
Yes I did try pinging from both VMs but packets were not reaching other VM
though I could see those packets (from both VM) in VPP trace hence I thought
VPP is dropping those packets(show error output was the basis). Anyways I will
try this again and try to debug deeper.
I also want to say here that sometime few packets did reach remote VM and ping
works on my ThunderX2 but that percentage is negligible. I see 99% packet loss
in ping statistics. 1% of them did work successfully and ping latency for those
1% packet is huge almost 6-7 min. For this i had to run ping for a longer
period to see these 1% successful ping.
Thanks,
Nitin
On 29-Sep-2017, at 06:33, Steven Luong (sluong)
<slu...@cisco.com<mailto:slu...@cisco.com>> wrote:
Nitin,
If you are referring to these counters in show error as transmission errors,
they are not really errors. Did you get a chance to try to send pings from the
other direction and capture the trace to see if the other VM is working fine? I
am able to ping VM -> VPP -> VM for both IPv4 and IPv6 without any problem on
my ThunderX.
[2KDBGvpp# sh err
Count Node Reason
17 l2-output L2 output packets
17 l2-learn L2 learn packets
3 l2-learn L2 learn hit updates
17 l2-input L2 input packets
17 l2-flood L2 flood packets
DBGvpp#
Steven
From: "Saxena, Nitin" <nitin.sax...@cavium.com<mailto:nitin.sax...@cavium.com>>
Date: Thursday, September 28, 2017 at 5:44 PM
To: "Steven Luong (sluong)" <slu...@cisco.com<mailto:slu...@cisco.com>>
Cc: "Narayana, Prasad Athreya"
<prasadathreya.naray...@cavium.com<mailto:prasadathreya.naray...@cavium.com>>,
"vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>"
<vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>>
Subject: Re: [vpp-dev] [vhost-user][armv8][v17.07.01] VM ping not working
properly
Hi steven,
I realised that ARP request not being resolved hence
I was capturing packet in remote VM using tcpdump.
I didnt see any packet on tcpdump. I feel packets are not reaching remote VM. I
am getting l2 forwarding errors and virtualinterface0/0/1-tx errors
Regards,
Nitin
On 28-Sep-2017, at 22:52, Steven Luong (sluong)
<slu...@cisco.com<mailto:slu...@cisco.com>> wrote:
Nitin,
Ping fails because the remote VM did not respond the ARP requests. As far as I
can tell, The ARP request packets were correctly sent from VirtualEthernet0/0/0
(11.0.0.1) to VirtualEthernet0/0/1 (11.0.0.2). But we are not getting any reply
back based on the output of show interface. Did you forget to configure the ip
address on the other VM? Please try ping from the other direction 11.0.0.2 ->
11.0.0.1 and capture the vhost-user trace again. I suggest you eliminate the
IPv6 traffic on the interface (if possible) to get a clean capture and “set
terminal pager off” to get a better show output on VPP.
Packet 17
00:05:25:459642: vhost-user-input
VirtualEthernet0/0/0 queue 0
[1m-- more -- (494-522/550)[0m
[2K virtio flags:
SINGLE_DESC Single descriptor packet
virtio_net_hdr first_desc_len 54
flags 0x00 gso_type 0
num_buff 0
00:05:25:459654: ethernet-input
ARP: 00:00:00:00:00:01 -> ff:ff:ff:ff:ff:ff
00:05:25:459665: l2-input
l2-input: sw_if_index 1 dst ff:ff:ff:ff:ff:ff src 00:00:00:00:00:01
00:05:25:459671: l2-learn
l2-learn: sw_if_index 1 dst ff:ff:ff:ff:ff:ff src 00:00:00:00:00:01 bd_index 1
00:05:25:459678: l2-flood
l2-flood: sw_if_index 1 dst ff:ff:ff:ff:ff:ff src 00:00:00:00:00:01 bd_index 1
00:05:25:459682: l2-output
l2-output: sw_if_index 2 dst ff:ff:ff:ff:ff:ff src 00:00:00:00:00:01 data 08
06 00 01 08 00 06 04 00 01 00 00
00:05:25:459687: VirtualEthernet0/0/1-output
VirtualEthernet0/0/1
ARP: 00:00:00:00:00:01 -> ff:ff:ff:ff:ff:ff
request, type ethernet/IP4, address size 6/4
00:00:00:00:00:01/11.0.0.1 -> 00:00:00:00:00:00/11.0.0.2
00:05:25:459694: VirtualEthernet0/0/1-tx
VirtualEthernet0/0/1 queue 0
virtio flags:
SINGLE_DESC Single descriptor packet
virtio_net_hdr first_desc_len 1536
flags 0x00 gso_type 0
num_buff 1
[1m-- more -- (522-550/550)[0m
[2KDBGvpp# sh err
Count Node Reason
17 l2-output L2 output packets
17 l2-learn L2 learn packets
3 l2-learn L2 learn hit updates
17 l2-input L2 input packets
17 l2-flood L2 flood packets
DBGvpp# sh int
Name Idx State Counter Count
VirtualEthernet0/0/0 1 up rx packets
23
rx bytes
1370
tx packets
11
tx bytes
866
drops
8
VirtualEthernet0/0/1 2 up rx packets
11
rx bytes
866
tx packets
23
tx bytes
1370
local0 0 down
DBGvpp#
From: "Saxena, Nitin" <nitin.sax...@cavium.com<mailto:nitin.sax...@cavium.com>>
Date: Thursday, September 28, 2017 at 9:26 AM
To: "Steven Luong (sluong)" <slu...@cisco.com<mailto:slu...@cisco.com>>
Cc: "Narayana, Prasad Athreya"
<prasadathreya.naray...@cavium.com<mailto:prasadathreya.naray...@cavium.com>>,
"vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>"
<vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>>
Subject: Re: [vpp-dev] [vhost-user][armv8][v17.07.01] VM ping not working
properly
Hi Steven,
Please find attached tarball having valid output for show vhost.
This time I did ping 11.0.0.2 -c 10 and captured 10 packets in trace.
Thanks,
Nitin
________________________________
From: vpp-dev-boun...@lists.fd.io<mailto:vpp-dev-boun...@lists.fd.io>
<vpp-dev-boun...@lists.fd.io<mailto:vpp-dev-boun...@lists.fd.io>> on behalf of
Saxena, Nitin <nitin.sax...@cavium.com<mailto:nitin.sax...@cavium.com>>
Sent: Thursday, September 28, 2017 9:32 PM
To: Steven Luong (sluong)
Cc: Narayana, Prasad Athreya; vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>
Subject: Re: [vpp-dev] [vhost-user][armv8][v17.07.01] VM ping not working
properly
Hi steven,
I did IPv4 ping. Some ICMPv6 router solicit messages were captured but those
were not part of IPv4 ping which I did.
I will send show vhost output soon.
Regards,
Nitin
On 28-Sep-2017, at 19:30, Steven Luong (sluong)
<slu...@cisco.com<mailto:slu...@cisco.com>> wrote:
Sexena,
Before you ping, type “show vhost” to make sure the interfaces have memory
region. You typed “show vhost” prior to the interface is ready. So, it does not
have useful information. Would you collect “show vhost” when the interface is
ready?
Also, you send ipv6 pings. Would you please also try ipv4 pings. I got it
running and pings went through on ThunderX last week with ipv4. I’ll try ipv6
later today to see if the problem is specific to ipv6.
Steven
From: "Saxena, Nitin" <nitin.sax...@cavium.com<mailto:nitin.sax...@cavium.com>>
Date: Thursday, September 28, 2017 at 1:23 AM
To: "vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>"
<vpp-dev@lists.fd.io<mailto:vpp-dev@lists.fd.io>>, "Steven Luong (sluong)"
<slu...@cisco.com<mailto:slu...@cisco.com>>
Cc: "Narayana, Prasad Athreya"
<prasadathreya.naray...@cavium.com<mailto:prasadathreya.naray...@cavium.com>>
Subject: Re: [vhost-user][armv8][v17.07.01] VM ping not working properly
Hi All,
I am running vhost-user on aarch64 SoC using vpp v17.07.01. I am running 4.12.9
kernel (ubuntu-16.04) both on host and VM. I am able to launch VM's
successfully but my ping between two VM's does not work. In show error output I
can see L2 forwarding errors in VPP. Attached tarball containing logs with
"debug vhost-user on" and "show trace"
I also tried adding following in my startup script.
vhost-user {
coalesce-frames 0
}
Any pointer will be helpful.
Thanks,
Nitin
_______________________________________________
vpp-dev mailing list
vpp-dev@lists.fd.io
https://lists.fd.io/mailman/listinfo/vpp-dev
