John Ballance <jwb@xxxxxxxxxx> wrote: > the iyo can deliver sharefs stuff far faster at 100mb than earlier > riscos machines can absorb the data. this results in the rpc taking all > its time responding to interrupts delivering data in and has no time > left to use the data. > > some work has being going on to make sharefs better in retries, but the > underlying issue IS that the rpc (etc) is slow and cannot be safely > spoken to much faster. > > protocols that do more handshaking (e.g. because they are designed to > travel much further than a local intranet) generally recover better. Or they don't need to recover in the first place. Anything using TCP, for example, will automatically get the transmission window size adjusted to match the connection speed. NFS over UDP sends a reply packet for every received request, and despite this extra overhead it is both faster and more reliable than ShareFS over a connection with mismatched speeds. As I haven't seen the ShareFS protocol specification, this is just speculation, however it seems to me that the problem is with the ShareFS protocol, and is unlikely to be fixed without modifying the protocol at both ends of the connection. The fact that the connection speeds are mismatched just exaggerates the problem. Alex -- Alex Waugh alex@xxxxxxxxxxxxx PHP, Roots, Subversion, WebJames and more from http://www.alexwaugh.com/ --- To alter your preferences or leave the group, visit //www.freelists.org/list/iyonix-support Other info via //www.freelists.org/webpage/iyonix-support