🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.sys.dec
Article #4513

Re: Cache parity error count for cpu

#4513
From: Max Maklin
Date: Fri, 29 Jan 1993 03:17
49 lines
3254 bytes
In article <1993Jan27.055647.25873@cs.aukuni.ac.nz> rob@cs.aukuni.ac.nz (Rob Burrowes ) writes:
:^()  I just installed ULTRIX 4.3 on a DECStation 5000/200. I am now getting
:^()  a large number of messages in the system log which state
:^()  "Cache parity error count for cpu 0 is _581"
:^()  next message changes the _581 to something else. The OS Event type is 250,
:^()  which the manual says is informational. The hardware self checks ok and
:^()  everything seems to be running fine. I am just getting a very large syserr log.
:^()  
:^()  Is this 4.3 or has the cache taken a dive and it is just a coincidence. A
:^()  5000/120 running 4.3 doesn't do this (it just logs the odd message with
:^()  cryptic numbers in the "MESSAGE" field) so I guess the hardware may be at fault.
:^()  
:^()  What is the "Cache parity error count" , which cache is it refering too, and
:^()  why is this not reported as a Hardware error rather than an informational
:^()  message?
:^()  
:^()  

Wow 581 is quite a bit.  When was the last time you rebooted this machine?
I will assume it has been a while.  I already posted about this about 6
months.  I had it escalated through support.  The story goes something
like this.

Some code that had been monitoring this for a while was enabled to write
error log records starting with Ultrix 4.2A.  This still happens to me
on my 5500 and 5000/200.  I get between 10-20/week depending how busy.
Should you be worried?  That depends on the first question.  If you
have rebooted in the last week, I would probably have DEC come take 
a look and maybe request a new CPU boards.  Ensure that you are at the
most recent FCO for the 200.  I think it is F03, but this is just from
memory.  Note, the error log is correct in informing you it as 250, since
it is a self-correcting event.  Performance-wise you do not want to
be getting too many in a short time.

The reason you probably do not see it on the newer models is that the
motherboard contains different chipsets (read no flaky Siemens chips).

Note: Please everyone refrain from requesting new boards (read never
      should have enables this message).  They are for the large
      majority harmless.  Besides given that there is less and less
      people around we need them concentrating their efforts on
      providing us Internet access to patches, right! :)

Max
-- 
Max Maklin                                                P.O. Box 9707 
Technical Analyst	                           3755 Riverside Drive
Cognos Incorporated                              Ottawa, Ontario CANADA
Internet: maklinm@cognos.COM                 UUNet: maklinm@cognos.UUCP
Technical Details
Message-ID: <1993Jan29.031717.27284@cognos.com>
Path: arc1-ca.pugleaf.net!sparky!uunet!UB.com!pacbell.com!ames!agate!spool.mu.edu!howland.reston.ans.net!zaphod.mps.ohio-state.edu!uunet.ca!cognos!maklinm
References: <1993Jan27.055647.25873@cs.aukuni.ac.nz>

Show Full Headers
Complete Article Headers
Newsgroups: comp.sys.dec,comp.unix.ultrix
Path: arc1-ca.pugleaf.net!sparky!uunet!UB.com!pacbell.com!ames!agate!spool.mu.edu!howland.reston.ans.net!zaphod.mps.ohio-state.edu!uunet.ca!cognos!maklinm
From: maklinm@cognos.com (Max Maklin)
Subject: Re: Cache parity error count for cpu
Message-ID: <1993Jan29.031717.27284@cognos.com>
Keywords: cache
Sender: news@cognos.com (News Administrator)
Nntp-Posting-Host: concorde
Organization: Cognos Incorporated, Ottawa CANADA
References: <1993Jan27.055647.25873@cs.aukuni.ac.nz>
Date: Fri, 29 Jan 1993 03:17:17 GMT
Lines: 49