On Fri, Jun 05, 2020 at 09:28:23AM +1200, Brian E Carpenter wrote: > > On the UDP side, there should be no issue. Anything above UDP has > > to tolerate loss anyway, so there should be mechanisms that will > > require from corruption-treated-as-loss. > > That depends. If a single bit inversion can get through undetected, > the application layer might never detect an anomalx. (See what I did > there?) A long time ago, back at Sun, we had a NIC on a server that would introduce paired-bit errors at offsets such that all checksums in the stack were defeated occasionally, which then corrupted Teamware back when Sun used Teamware over plain NFS for its source repositories(!). IIRC there was a nifty blog entry about this, but I'm not sure how I'll ever find it. Anyways, that one achieved legendaray status at Sun. It was very hard to find, much harder than your typo :) Nico --
