djh@cs.mu.oz.au (David Hornsby) (09/06/89)
We have just had an interesting problem with our ImageWriter IIs on LocalTalk. The symptoms were that the printer would curl up its toes and die every few minutes, sometimes taking the whole local net with it (by producing a continuous stream of CTS packets). It was not visible in the Chooser nor would it accept jobs. The only solution was to power cycle it - only to have it die again minutes later. Not nice. My first response was to junk the printers, but that is unfortunately impractical. The culprit was found to be the following packet which was being produced regularly by our gateways ... ff b5 01 00 39 48 48 16 45 00 00 34 05 31 00 00 3c 11 0c 73 80 fa b5 81 80 fa b5 9f 02 08 02 08 00 20 8d 77 01 01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 10 For the curious, this packet is a perfectly legal IP-inside-shortDDP broadcast RIP request (type 1, version 1, null tuple) [yes, our gateways implement RIP]. Of more interest, type 2 RIP packets (RIP_RESPONSE) don't affect the ImageWriter in any way. The packets are similar (except for the length bytes) until at least the 12th byte. Questions: 1. What the hell is an ImageWriter doing looking at packets sent to socket 0x48 of a type (0x16) way outside the Apple restricted range ? The particular socket or type doesn't even matter, I have varied both with no change in behaviour. 2. There are rumours about a new ROM set for the AppleTalk card for ImageWriters. Could some authoritative Apple type please comment on this. What other problems (IE: incorrect NBP response to DDP header address) are fixed ? We have the same problems with printers purchased two years apart. The ROM has "341 0034 A" written on it. 3. If anyone has ImageWriters with new ROMs, could they please fire this packet at it and let me know the result. The "final" solution ... Well, it seems that if you append 4 more null RIP tuples to the packet then the ImageWriter is perfectly happy. Our gateways now produce these bogus packets only on their LocalTalk ports and everyone (?) is happy. Regards, David Hornsby ...!uunet!munnari!djh Professional Officer djh@munnari.OZ.AU Department of Computer Science +61 3 344 4044 The University of Melbourne, Parkville, 3052, Victoria, Australia.