[j-nsp] tfeb0 LUCHIP(0) PPE_15 Errors sync xtxn error

MSusiva ssiva1086 at gmail.com
Sat Jul 19 10:56:28 EDT 2014

Yes, this is related to a sampling. You may see a ppe trap file generated
on the router that would give more information to understand the cause of
this issue.
I would suggest you to work with JTAC on this.

Thank You,

Date: Tue, 15 Jul 2014 15:17:23 +0300
From: Saku Ytti <saku at ytti.fi>
To: juniper-nsp at puck.nether.net
Subject: Re: [j-nsp] tfeb0 LUCHIP(0) PPE_15 Errors sync xtxn error
Message-ID: <20140715121723.GA9724 at pob.ytti.fi>
Content-Type: text/plain; charset=us-ascii

On (2014-07-15 11:25 +0400), dmitry at zhigulinet.ru wrote:


> Jul 15 09:00:38   tfeb0 LUCHIP(0) PPE_15 Errors sync xtxn error
> Jul 15 09:00:38   tfeb0 PPE Sync XTXN Err Trap:  Count 50, PC 617c,
0x617c:  flow_age_read_active_timestamp     0x617c:  flow_age_ipv4

Each LUCHIP has 16 PPE, lookup tasks are sprayed to PPEs to distribute work,
each PPE has 20 contexts, so that PPE can work even when doing I/O wait.

You can get more data about PPE 15 from 'show luchip 0 ppe 15 ...' maybe
'trap_info', after logging to the TFEB via 'start shell pfe network tfeb0 '

> What is it?

I'd guess like Abhi that it's related to IPFIX. I don't think it has
to do with linecard synchronization like Iftikhar suggested, PPE does not
really experience or talk to linecard.

There is absolute limit how long PPE can work, maybe 'sync xtxn' implies PPE
worked longer than allowed?
It does not look too bad to me really, if it's failing to read flows active
timestamp, does not seem the reprecussions would be too dramatic.


More information about the juniper-nsp mailing list