The last processor I worked on/with that had memory problems, single bit, double bit errors, etc was on a System 370 mdl. 135.
We were running DOS at the time and the memory errors would occassionally put the cpu into a hardwait. On one occurance the box
actually IPL'd itself. The CE said he had heard that this model could (as in may be able to) do this but he had never
personally witnessed it. I know this happened because I was the operator at the time, was in the computer room, but no where near
the console. The IPL diskette drive, top left on operartor panel, started spinning and clicking. Most operators know the normal sounds
in a computer room and when I heard that noise, I was wondering what it was. At about the time I located the source, 2 managers
came into the computer room and wanted to know what had happened. I told them exactly what happened, they had their doubts,
but they knew I wasn't a person to lie. Only after they talked to the CE did they fully believe my story.
I know, a long winded story, that only needed about 3 sentences.
Regards.
Steve Gentry
| Colin Allinson <[EMAIL PROTECTED]>
Sent by: The IBM z/VM Operating System <[email protected]> 05/10/2006 04:57 AM
|
To: [email protected] cc: Subject: Memory Reliability |
In all my years working with IBM mainframe systems I have never seen a memory failure that caused recognisable problems.
My understanding is that the internal redundancy, hardware checking and auto correct make such an event very unlikely - but how unlikely? Has anyone ever experienced such a problem?
If a problem did occur at the level that affected VM would it be more likely to bring VM down or just affect a specific user referencing the memory blocks with the problem ?
Regards
Colin Allinson
