Conference mvblab::alphaserver_4100

  I have just read the notes entry and a similar event occurred by BMW in
Germany, but unfortunately the note does not give aan explanation.
here is what happened by the reboot-

 Customer rebooted Alpha and console hung . He managed to reboot the machine
remotely by using RCU and issuing a Reset.He has rebooted the machine again
this week and the machine rebooted without problem. But the customer would like
an explanation and guarantee it will not regulaely happen


        03-Mar-1997 21:24:02
03-Mar-1997 21:24:32
03-Mar-1997 21:24:32  halted CPU 0
03-Mar-1997 21:24:32
03-Mar-1997 21:24:32  halt code = 5
03-Mar-1997 21:24:32  HALT instruction executed
03-Mar-1997 21:24:32  PC = ffffffff800521dc
03-Mar-1997 21:24:32
03-Mar-1997 21:24:32  CPU 0 booting
03-Mar-1997 21:24:32
03-Mar-1997 21:24:32  (boot dkd100.1.0.4.0 -flags 0,0)
03-Mar-1997 21:24:32  �Building FRU table
03-Mar-1997 21:24:32  block 0 of dkd100.1.0.4.0 is a valid boot block
03-Mar-1997 21:24:32  reading 1004 blocks from dkd100.1.0.4.0
03-Mar-1997 21:24:32  bootstrap code read in
03-Mar-1997 21:24:32  base = 200000, image_start = 0, image_bytes = 7d800
03-Mar-1997 21:24:32  initializing HWRPB at 2000
03-Mar-1997 21:24:32  initializing page table at 1f2000
03-Mar-1997 21:24:32  initializing machine state
03-Mar-1997 21:24:32  setting affinity to the primary CPU
03-Mar-1997 21:24:32  configuring I/O adapters...
03-Mar-1997 21:24:32    ncr0, hose 1, slot 1, bus 0
03-Mar-1997 21:25:03    ncr1, hose 1, slot 3, bus 0
03-Mar-1997 21:25:03    pfi0, hose 1, slot 4, bus 0
03-Mar-1997 21:25:03  Unable to malloc memory for Decompression for NET
03-Mar-1997 21:25:03  !!! Console entry context is not valid - Reset the system
!!!
03-Mar-1997 21:25:03
03-Mar-1997 21:25:03  Brk 0 at 0005FD0C
03-Mar-1997 21:25:03
03-Mar-1997 21:25:03  0005FD0C ! BPT

    
    The same problem is discussed in note 1115 of the
    msbcs::turbolaser conference.  Any chance we have a 
    similar situation with our SRM?
    
    Dennis
    

    From the "Building FRU table" line in .3, I am guessing that the
    version of console on this machine is V2.0-3 or less. We have made some
    improvements since then in terms of console memory heap reclamation. I
    would recommend upgrading the console on this machine. In the future,
    please always post console and operating system version when reporting
    problems.

    Note that even the most recent V4.8 SRM console has some small memory leaks
    in certain drivers (e.g. 3KB in the KZPSA driver, 1KB in memory channel
    init code). Until these leaks are fixed in a future release, over time, and
    depending upon system configuration, large numbers of reboots without a
    periodic system reset will result in sufficient heap loss such that the SRM
    will be unable to decompress/malloc a given DLL (e.g. network driver and
    protocol layers) resulting in the type of error described in .3.
    
    One customer was rebooting every night - for whatever reason. He had
    several KZPSAs and ran into the "unable to malloc" problem. Our suggestion
    was to periodically reset the system during reboots until the console issue
    can be resolved.

    FWIW, the memory leaks that we were aware of have all been fixed as of
    V4.8-5 SRM console and beyond.