stevens@sri-unix (10/28/82)
We have had the machine check 2 problem with two 11/750's running 4.1bsd. We called DEC last week and they say that the problem is known but it is not an official ECO. Apparently they have used two different manufacturer's chips on one of the CPU boards and one of the two is "noisier" than the other. They are coming out to check both of our systems and will replace the chip if it is from the "noisy" source. (Our local DEC field service tracked this down within DEC and found someone who was "very knowlegable about UNIX"). Changing the board with another containing the same chip will probably not solve anything. Also, the October 1982 "*UNIX softalk" (a new rag that I got unsolicited) from "international data services" (408-730-unix) has some info on this and they claim to have a software patch for 4.1bsd. They have sent me the info, but I havent received it yet. Rich Stevens, Kitt Peak National Observatory
charliep@tekgds.UUCP (Charles E. Perkins) (09/16/83)
We have had many crashes on our VAX 11/750 system running UNIX 4.1bsd. After a few times we got tired of calling DEC service out; they replaced the same board several times and it has not seemed to improve things. Is this a common problem? Are there known reasons for this problem? Is it somehow caused by software? My current theory is that DEC has some wierd problem that shows up under UNIX but not VMS. If it was really a software problem, why does the Remote Diagnostic Center always conclude otherwise (it invariable reports a problem with the translation lookaside buffer, hence the board replacement). I would certainly appreciate any comments you might have about this (especially solutions or bug fixes!) Even if your machine runs perfectly, that is good information, because then I can tell DEC that they need to get us some better hardware like other sites have. (!) Usually this seems to happen under moderately heavy use which for us means a load average of over 4. We have root and /usr on a RA80, and the rest of our file systems on a RA81. Thanks for your help. Charles Perkins Logic Design Systems PS. Has anybody written a dump routine for UDA50 disks?