• Tag : hardware


Optimizing hardware utilization for unused (S,G) routes

Description Current EOS PIM sparse mode implementation installs an (S, G) route in the hardware whenever it receives an (S, G) join from the PIM neighbor. Prior to using this feature, devices could have higher hardware utilization because EOS installs all the routes corresponding to the (S, G) joins. This is especially evident if the devices peers with other autonomous system receivers which are administered by an external source. These receivers can send a large number of (S, G) joins for invalid sources leading to exhaustion of hardware resources. With the support of this feature, PIM sparse mode doesn’t install...
Continue reading →

Enterprise Internet Routing

Overview The objective of this document is to cover the most common Enterprise Internet Routing use case. The information provided here is based on two Arista switches peering with two ISP’s (Internet Service Providers) for redundancy. There are many other valid deployment models that are not covered in this document. Terminology BGP – Border Gateway Protocol ISP – Internet Service Provider BGP Peering – a session between two BGP routers that allows exchange of routes Full Internet Routing Table – all public routes on the Internet AS – Autonomous System – defines domain boundaries IGP – Interior Gateway Protocol EGP...
Continue reading →

Reversing The Airflow of a Running Switch

Occasionally customers have a switch in production that they need to change the airflow direction of e.g. from a front-to-rear (-F) to a rear-to-front (-R), or vice-versa, without shutting down the switch. The following procedure outlines the steps to follow in order to accomplish this. This procedure assumes that you already have the replacement fans and PSUs on hand and are ready to perform the swap. Start with the switch powered up, both power supplies powered and providing power to the switch. Gain access to the switch’s serial console to check status and run CLI commands as need be. Add...
Continue reading →

System log shows switch interfaces entering then leaving promiscuous mode; cpu error and reload days later

Hello,   We had a switch error message recently and was wondering if anyone else had a similar experience. The kernel log shows the switch interfaces entering promiscuous mode a little over two weeks prior to the cpu error. This was a packet capture done to diagnose an attached server. We have several other Top of Rack switches running the same software version without issue.   Has anyone else seem similar? Searching for these kernel dump errors has not turned up much Arista specific discussion, so that’s the reason for this post. Hopefully the question and/or answer can help someone...
Continue reading →


Get every new post on this blog delivered to your Inbox.

Join other followers: