[j-nsp] One or more stream blocked

[Darrell Hyde]

> Any info on whether or not the above constitutes a symptom (of
> hardware failure or misconfiguration) would be appreciated.  None of
> our other MXes behave this way.

The best I've been able to get out of JTAC so far is that this is
evidence of the ICHIP being in a wedge condition. Google details on
what this means are slim. JTAC indicates this refers to a condition
where packets enter an interface to be forwarded and are never
switched to their next-hop.

There was an AI script release last year that refers to an "i-chip wedge issue":

http://www.juniper.net/techpubs/software/management/ais/ais21/rn-ai-scripts-21/rn-ai-scripts-21.pdf

The bug is documented in PR277853, and is fixed in the release we're running:

COMMITTED-IN RELEASE	8.4R4; 8.5R3; 9.0R3; 9.1R1; 9.2R1; 9.3R1

When packets are queued for several seconds because of interface
congestion, packet CRC errors are reported in some cases. In other
cases, the egress interface stops forwarding traffic (either all
traffic is halted or only packets larger than 320 bytes are not
forwarded). As a workaround, configure queues with a transmit rate of
at least one percent of the line rate. When using "strict-high"
priority queues, include policers to prevent interface congestion and
the starving of lower-priority queues. Another alternative is to use
"high" priority queues instead of "strict-high" priority queues.

I'll share whatever JTAC comes up with.

D.