> When class-of-services classifiers are configured, are they setting the loss-priority and code-point or are the matching them?
They are actually setting the loss-priority bit. There is no loss-priority bit in MPLS packets, so it is up to the classifier to assign one on the basis of the EXP marking.
Please check this:
http://www.juniper.fr/techpubs/software/junos/junos91/swconfig-cos/setting-the-plp-with-a-ba-classifier.html
So if you write a statement like this:
set class-of-service classifiers exp BA_CORE_EXP_CLASSIFIER forwarding-class NC loss-priority low code-points 111
you are actually requesting all MPLS traffic which has been tagged with EXP-bits 111 to be assigned the forwarding class NC, and a loss priority of low.
You would do this (set loss-priority to 'low') to make sure your network control traffic is never discarded.
Also, there is a bit of confusion when it cames to these two statements:
> class-of-service rewrite-rules exp CORE_EXP_REWRITE forwarding-class NC loss-priority low code-point 111
> - Egress MPLS exp NETWORK CONTROL packets with a low loss-priority and a code-point of 111 will be rewritten with a low loss-priority and a code-point of 111 (no change).
It is actually the other way around.
You are requesting MPLS traffic which has been classified as belonging to the NC class with low loss-priority to be rewritten with EXP-bits 111 when leaving the router.
Other traffic in the same NC class but with loss-priority of 'high' would not be affected by this rewrite rule.
About the loss-priority flag, it is mostly used to have separate drop profiles, i.e. to prioritize traffic within the same queue.
Hope it helps..
Saverio