On Sat, Sep 15, 2012 at 10:10 PM, Paul Graydon <p...@paulgraydon.co.uk>wrote:
> On 09/14/2012 09:40 PM, Phil Pennock wrote: > ... > >> For instance, a dirty cable with noise causing corruption might cause >> checksums to fail and retransmissions, but the link is up and in use. >> Or the device on the far side might be wedged such that it keeps the >> link "up" with protocol management reporting health but doesn't actually >> forward packets. >> > This is more what I'm thinking about though. Is there any resiliency > stuff built in to link aggregation. e.g. if you're having to do a lot more > re-transmissions of packets down a particular path, maybe reduce or > eliminate the path from use? (all the while there are sufficient additional > ports, naturally) > TCP retransmissions are handled by the end points and the local link has no idea that they are occurring. Ethernet checksum errors, on the other hand, are detectable by the receiver so it could unilaterally drop a link when it sees too many bad packets. Hopefully the send side would notice and drop that link from its active set as well. I don't know the intricacies of the various link aggregation methods well enough to know if anybody actually does this.. Based on this thread, it would seem that if they do it doesn't always work the way one would like. Bill Bogstad
_______________________________________________ Tech mailing list Tech@lists.lopsa.org https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech This list provided by the League of Professional System Administrators http://lopsa.org/
