Conference mvblab::alphaserver_4100

577.0. "IOD's Failing Power UP Selftests" by TAINO::JFUSTE (JOSE FUSTE) Sat Apr 26 1997 13:26

    The following problem is happening to us in a new system;
    
    	When the system was power up for the first time after received at
    the customer site the IOD always fail the POwer up Selftest. The OCp
    message is as follows;
    			P0 FAIL 1 IOD0
    			P0 FAIL 1 IOD1
    I was involved in the problem yesterday. The engineer onsite has been
    working with the system for the past two weeks. He update the firmware
    to the latest included with the system  in floppies because the CPU's
    are the new B3004-AA 466Mhz. After the console came up shows thath both
    IOD's failed the power up selftests. He replaced the B3040 twice, the
    B3050 and the CPU & Memory motherboard. I was helping him out yesterday
    and seems to me that the problem looks like a corrupted SRM all the
    time. This is a Rackmount system and another dodge box came in the same
    cabinet but this one is working good. We check all the envairomentals
    between both systems and there is no difference. I ran the IOD diags 
    and always fail with all the registers in 0's. Also when running LFU
    sometimes the system hung forever and need to be rebooted. The memory
    test were put in "full" and they always pass. We get rid of all the PCI
    options to work with the system at a minimun level only with the VGA
    card. We also called support but the person who answer doesn't had a
    clue. 
    
    Any input will be helpful.
    
    Thanks,
    	Jose

    
    	How much and what type of memory?  How many CPUs, what rev?
    	If possible, can you post console sho config, sho fru, and sho *?
    
    	- Walt
    

    Thanks for answering my note. I finally found the problem. The problem was
    the new memory MS330-GA 2gb. The system was reporting IOD failure and
    all the memory tests were passing without any single error. I suspect
    the memory because no matter what we try to do in the console most of
    the time the console hung. Also, I noticed in the OCP that memory test
    24 was acting different with the bad memory. When you set the EV
    memory_test to full it takes longer and the display shows the following
    symbols "+ - / \" going in circles. With the bad  memory the "/" symbol
    always stay. Thanks again.
    
    Jose