camargo@cs.columbia.edu (Francisco Camargo) (11/27/89)
This message was posted during the hollidays . I believe that not many of you saw it. So, here it goes again.... I'm having such a difficult time figuring out this problem that I decided to cry for help. In a company where I perform occasional system maintenance, we have Advanced Netware 2.15 NON-DEDICATED running on a Compaq 386/25 w/3000Mb hard-disk and 2 NOVELL NE-2000 Ethernet boards. Also, we used to have a tape-backup driver in the file server, but it was removed after advise by the Novell people, this without solving the problem, which I now describe. The problem: Once every one or two weeks, the file server goes down with the error message INVALID TASK ABEND. DMA Error XXXXXXX (some weird code). The Novell error messages manual indicates that this message can be related to a bad DMA chip in the ethernet boards, but the NE-2000 boards don't have a DMA chip, nor they use the DMA in the motherboard. In fact, they don't support DMA at all. The next reference in the error message manual is for "Power Problems", bad line conditioning and so on. Well, the system is being powered by an Elgar 1100, which has a line conditioning and battery backup support. In fact, we replaced it 3 times hoping to solve the problem. It didn't help. Talking to the people at Novell, they reported a problem with the NE-2000 boards manufactured prior to August 14/89, which had a bad DRAM chip and caused an intermitent problem very hard to detect. We replaced the boards, but the problem keeps occuring. In fact, we can track it down to some interference with the TAPE BACKUP driver. Originally, it was installed in the file server, but has been moved to a workstation. Whenever somebody inserts a tape into the unit, runs a backup, and then removes the tape, the workstation goes blank, and sometimes the file server goes down as well. Actually, this doesn't happens all the time, but with an annoying frequency to cause us concern about loosing data. BTW, all interrupts and I/O address have been checked, and there is no conflict there. In fact, other than this occasional problem, the network runs pretty well, with a more than adequate performance. We had the file server fully diagnosed, passing all tests, the power lines checked, the NE-2000 boards replaced (by newer NE-2000), 3 different ELGAR units tested, but the problem won't go away. Has anybody ever experienced such problem ? Is there anything about the NE-2000 boards or the ELGAR power backup unit which makes them not recommended ? Can anybody point to any problem in the Compaq Tape Backup drivers (Running SY-TOS) which causes interference with Novell ? Any light in this problem will be greatly appreciated. camargo@cs.columbia.edu