[j-nsp] Frame loss during Ethernet test

Gökhan Gümüş ggumus at gmail.com
Tue Jan 24 08:23:11 EST 2012


Dear Tim,

Thanks for the information.
Routers are MX and software version is same at both end which is 10.0R4.7
I also checked the syslog messages and saw only repeated cosmetic messages
as follows;

Jan 24 12:18:12 lon01 /kernel: PCF8584(WR): busy at start, attempting to
clear
Jan 24 12:18:12  lon01 /kernel: PCF8584(WR): (i2c_s1=0x08, group=0xf,
device=0x54)
Jan 24 12:26:02  lon01 /kernel: PCF8584(WR): transmit failure on byte 1
Jan 24 12:26:02  lon01 /kernel: PCF8584(WR): (i2c_s1=0x00, group=0xf,
device=0x54)
Jan 24 12:32:33  lon01 /kernel: PCF8584(RD): ack failure on 2nd last byte
Jan 24 12:32:33  lon01 /kernel: PCF8584(RD): (i2c_s1=0x80, group=0xe,
device=0x54)
Jan 24 12:32:33  lon01 /kernel: PCF8584(WR): busy at start, attempting to
clear

Could there be a bug also in this software which causes those frame loss?
Also, i have one question. Even if we see frame loss during test, could it
be possible to use the circuit in production without any problem?

Thanks and regards,
Gokhan Gumus


On Tue, Jan 24, 2012 at 2:13 PM, Tim Jackson <jackson.tim at gmail.com> wrote:

> This isn't running on an MPC carded MX running 10.2 code is it?
>
> There was a bug that bit us that had some random small bits of frame
> loss on CCC circuits on MX80 on 10.2.. I can't recall the bug ID, but
> it was fixed in 10.4R6.. If I remember right, check your logs and
> you'll see some parity errors scrolling through when there is frame
> loss.
>
> --
> Tim
>
> On Tue, Jan 24, 2012 at 4:03 AM, Gökhan Gümüş <ggumus at gmail.com> wrote:
> > Dear all,
> >
> > One of our customer complaining a small amount of frame loss on their
> > service.
> > I see that lost packets when i compare input and output statistics on the
> > interface.
> > We are terminating service with ccc configuration.
> > At remote end, we have a hardware loop and customer connects its tester
> to
> > A-end sends traffic.
> > When we run the test for 2-3 mins, we have no frame loss but at longer
> test
> > we start to see 10-20 frame loss.
> > Frames are lost when our PE puts sent test traffic into corrresponding
> LSP
> > but there is no issue in our LSP.
> >
> > I do not know the exact principle of these testers but i checked all the
> > possible reasons which may cause frame loss
> > Delay, errors on the line, oversubscription, etc...None of them is an
> issue
> > during test.
> >
> > Is there anybody who has such an experience with Ethernet testers?
> >
> > Thanks and regards,
> > Gokhan
> > _______________________________________________
> > juniper-nsp mailing list juniper-nsp at puck.nether.net
> > https://puck.nether.net/mailman/listinfo/juniper-nsp
>


More information about the juniper-nsp mailing list