[j-nsp] Strange latency when using traceroute.

[Stephen Fulton]

Hi Alex,

My first thought was BGP, since we're using mostly Cisco gear, but even 
with BGP having less than 50 routes (in a lab), I saw the issue.  Since 
the J6350 has not yet hit the full production stage, I haven't had a 
chance to really evaluate the CPU usage, but even now the CPU is idling 
between 90% - 93% most of the time.  Load is negligible though.

-- Stephen


Alex Campbell wrote:
> I have noticed this on J4350s as well (the J4350s are hop 3 and 4):
> 
>   1    <1 ms    <1 ms    <1 ms  vpn-aggr1-117gre.dtdesign.com.au
> [202.92.250.50]
>   2    <1 ms    <1 ms    <1 ms  fe-0-0-0.core1.117gre.dtdesign.com.au
> [202.92.250.1]
>   3     10 ms     9 ms     9 ms  ge-0-0-3.border1.190cit.dtdesign.com.au
> [202.92.251.157]
>   4     7 ms     9 ms     9 ms  ge-0-0-2.border2.190cit.dtdesign.com.au
> [202.92.251.18]
>   5     1 ms     1 ms     1 ms  as65000.border2.190cit.dtdesign.com.au
> [218.100.1.25]
>   6     1 ms     1 ms     1 ms  you-want-fries-with-your.ausix.net
> [218.100.1.14]
>   7    15 ms    13 ms    14 ms  pos7-0-0.switch1.syd.pacific.net.au
> [210.23.140.181]
>   8    13 ms    13 ms    13 ms  wasted.pacific.net.au [61.8.0.55]
> 
> We occasionally see spikes up to around 300 ms but I presume this is
> just the BGP scanner running.
> 
> Also the idle CPU usage on the J4350s seems strangely high - around 10%
> with almost no traffic going through the boxes.
> 
> 
> 
> -----Original Message-----
> From: juniper-nsp-bounces at puck.nether.net
> [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of Stephen Fulton
> Sent: Sunday, 1 July 2007 3:05 PM
> To: juniper-nsp at puck.nether.net
> Subject: [j-nsp] Strange latency when using traceroute.
> 
> Hi all,
> 
> I've noticed that when traceroute passes through our J6350 (using
> 8.3R15), that the returned latency is considerably higher from the
> router.  Pinging the router or devices past it are fine, but this is one
> of those things I know I'll be asked about, and I haven't been able to
> figure out.  Here's the redacted output from a sample traceroute:
> 
> [sfulton at optics ~]$ traceroute 192.168.254.2 traceroute to 192.168.254.2
> (192.168.254.2), 64 hops max, 40 byte packets
>   1  router1 (192.168.110.74)  0.337 ms  0.143 ms  0.494 ms
>   2  j6350 (192.168.98.65)  25.119 ms  28.051 ms  29.982 ms
>   3  router3 (192.168.254.2)  1.079 ms  0.743 ms  0.962 ms
> [sfulton at optics ~]$
> 
> Now there are filters protecting the routing engine, but I've removed
> all others for the purpose of testing.  CPU load is practically zero,
> and the J6350 is running OSPF and BGP.
> 
> Any ideas?
> 
> -- Stephen.
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp