Mailing List Archive

nprobe + zc troubles
I currently have zbalance_ipc configured with two instances per cluster id.

Normally, one would access the virtual device as “zc:10@0” and “zc:10@1”.

I have n2disk capturing off of zc:10@0 instance. Unfortunately, when
trying to run nprobe against zc:10@1 instance it only occasionally works.
When it fails, I get the following error:

16/Mar/2016 19:42:24 [nprobe.c:5348] Using packet capture length 9000
16/Mar/2016 19:42:24 [util.c:4344] ERROR: Cannot get hw addr for zc:10@1
16/Mar/2016 19:42:24 [pro/pf_ring.c:356] Initializing PF_RING socket on
device zc:10@1..
16/Mar/2016 19:42:24 [nprobe.c:5424] ERROR: Unable to open interface
zc:10@1.
16/Mar/2016 19:42:24 [nprobe.c:7080] ERROR: Unable to open interface zc:10@1
(zc:10@1: No such device exists (SIOCGIFHWADDR: No such device))
16/Mar/2016 19:42:24 [nprobe.c:7082] ERROR: Try using -i none if you do not
want capture from a NIC

The fix seems to be a complete restart of the pf_ring service.

Any advice on what to look at next?
Re: nprobe + zc troubles [ In reply to ]
Hi Jeremy
what do you mean with “occasionally” exactly? Is it just the "first time" perhaps?
It could be nprobe does not detaches the queue correctly on shutdown (crash or similar),
and it is no more able to reattach to the same queue until you restart zbalance_ipc.
Could this be the case?

Alfredo

> On 16 Mar 2016, at 20:48, Jeremy Ashton <jeremy.ashton@shopify.com> wrote:
>
> I currently have zbalance_ipc configured with two instances per cluster id.
>
> Normally, one would access the virtual device as “zc:10@0” and “zc:10@1”.
>
> I have n2disk capturing off of zc:10@0 instance. Unfortunately, when trying to run nprobe against zc:10@1 instance it only occasionally works. When it fails, I get the following error:
>
> 16/Mar/2016 19:42:24 [nprobe.c:5348] Using packet capture length 9000
> 16/Mar/2016 19:42:24 [util.c:4344] ERROR: Cannot get hw addr for zc:10@1
> 16/Mar/2016 19:42:24 [pro/pf_ring.c:356] Initializing PF_RING socket on device zc:10@1..
> 16/Mar/2016 19:42:24 [nprobe.c:5424] ERROR: Unable to open interface zc:10@1.
> 16/Mar/2016 19:42:24 [nprobe.c:7080] ERROR: Unable to open interface zc:10@1 (zc:10@1: No such device exists (SIOCGIFHWADDR: No such device))
> 16/Mar/2016 19:42:24 [nprobe.c:7082] ERROR: Try using -i none if you do not want capture from a NIC
>
> The fix seems to be a complete restart of the pf_ring service.
>
> Any advice on what to look at next?
>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
Re: nprobe + zc troubles [ In reply to ]
I was able to just replicate the issue by starting nprobe up and stopping
after a few seconds. I did this 6 times and then it failed again.

Restarting zbalance_ipc did not correct the issue.

On Wed, Mar 16, 2016 at 4:02 PM, Alfredo Cardigliano <cardigliano@ntop.org>
wrote:

> Hi Jeremy
> what do you mean with “occasionally” exactly? Is it just the "first time"
> perhaps?
> It could be nprobe does not detaches the queue correctly on shutdown
> (crash or similar),
> and it is no more able to reattach to the same queue until you restart
> zbalance_ipc.
> Could this be the case?
>
> Alfredo
>
> > On 16 Mar 2016, at 20:48, Jeremy Ashton <jeremy.ashton@shopify.com>
> wrote:
> >
> > I currently have zbalance_ipc configured with two instances per cluster
> id.
> >
> > Normally, one would access the virtual device as “zc:10@0” and “zc:10@1
> ”.
> >
> > I have n2disk capturing off of zc:10@0 instance. Unfortunately, when
> trying to run nprobe against zc:10@1 instance it only occasionally
> works. When it fails, I get the following error:
> >
> > 16/Mar/2016 19:42:24 [nprobe.c:5348] Using packet capture length 9000
> > 16/Mar/2016 19:42:24 [util.c:4344] ERROR: Cannot get hw addr for zc:10@1
> > 16/Mar/2016 19:42:24 [pro/pf_ring.c:356] Initializing PF_RING socket on
> device zc:10@1..
> > 16/Mar/2016 19:42:24 [nprobe.c:5424] ERROR: Unable to open interface
> zc:10@1.
> > 16/Mar/2016 19:42:24 [nprobe.c:7080] ERROR: Unable to open interface
> zc:10@1 (zc:10@1: No such device exists (SIOCGIFHWADDR: No such device))
> > 16/Mar/2016 19:42:24 [nprobe.c:7082] ERROR: Try using -i none if you do
> not want capture from a NIC
> >
> > The fix seems to be a complete restart of the pf_ring service.
> >
> > Any advice on what to look at next?
> >
> >
> > _______________________________________________
> > Ntop-misc mailing list
> > Ntop-misc@listgateway.unipi.it
> > http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>