Sorry, here is the updated gist link:
https://gist.github.com/ironpillow/a57cf283ce30ac344ef39e5425a81260
On Mon, Jan 8, 2018 at 3:02 PM, Carlito Nueno wrote:
> Hi all,
>
> I followed the home gateway tutorial [1] to setup a router on ubuntu box.
>
> I have a switch with two VLANs, vlan20 and vla
Hi all,
I followed the home gateway tutorial [1] to setup a router on ubuntu box.
I have a switch with two VLANs, vlan20 and vlan30, on port 1 connected
to VPP router box on port 1. So, its only one physical cable.
My current setup diagram and vpp config files:
https://gist.github.com/anonymous/
Hi,
Quick question, am I right in thinking that if I want to add FM10K VF support
in VPP I only need to modify
/src/plugins/dpdk/device/dpdk.h
/src/plugins/dpdk/device/format.c
/src/plugins/dpdk/device/init.c
Or am I missing something in there?
Thanks,
Dan B
_
Hi,
I followed Dave Barach’s instructions as mentioned in one ofthe posts,
To use the built-in leakfinder: build TAG=vpp_debug, set
#defineMHEAP_HAVE_SMALL_OBJECT_CACHE to 0 in .../src/vppinfra/mheap_bootstrap.h.
Then:
* Start vpp, configure it, and so forth.
* “memory-
Hi, I followed Dave Barach’s instructions as mentioned in one of the posts, To
use the built-in leakfinder: build TAG=vpp_debug, set #define
MHEAP_HAVE_SMALL_OBJECT_CACHE to 0 in .../src/vppinfra/mheap_bootstrap.h. Then:
* Start vpp, configure it, and so forth. * “memory-trace on” *
Hi Alberto,
Glad it helped. Concerning the new issues you’ve hit: are you running ATS with
multiple workers, i.e., does it fork multiple times like nginx does? If yes,
that may be the source of the problem for vcl. Dave (CC’ed) can provide more
details and I believe he’s working on a solution
Hi Florin,
Thanks for the hint. ATS was actually trying to bind to some ip addresses that
were not in the Fib.
Unfortunately, I have some other errors (that maybe recognize as well). The
output of ATS is now:
[11899] vcom_init...done!
[11899] vcom_constructor...done!
traffic_server: using root
Danny,
Thank you for your service to the community. I hope you will continue to
participate in the future, and I hope our paths will continue to cross :)
You are correct on the procedure, the TSC does not need to approve PTL
elections, though it would be courteous to inform the TSC, and do make s
Bump...
Would it make sense to talk about this some at the vpp-dev meeting tomorrow?
Ed
On Fri, Dec 22, 2017 at 2:14 PM Burt Silverman wrote:
> Search is somewhat painful with mailman. Otherwise I like it. Having said
> that, I have just learned about zgrep, and that simplifies the process,
>
Mheap.c has its own highly accurate memory leak checker. I haven’t tried the
valgrind integration in many years. Valgrind makes vpp run slowly enough to
make it unusable.
To use the built-in leakfinder: build TAG=vpp_debug, set #define
MHEAP_HAVE_SMALL_OBJECT_CACHE to 0 in .../src/vppinfra/mhea
Hi,
Vppctl shows that "rx out of buffer" incrementing even in a very low
rate 1Mpps .
We see somethings negative values ?!
What "rx out of buffer" indication stands for ? Which buffers ?
In the DPDK we see no indication for missing buffers ?!
-Shacha
Hi,
I want to check memory leakage on VPP so I use Valgrind memcheck tool to do
it,
but it gives me a lot of errors on output, like "Invalid read of size ..."
and
at the end of the valgrind memcheck report is saying :
"This is usually caused by using VALGRIND_MALLOCLIKE_BLOCK in an
inappropriate wa
12 matches
Mail list logo