[SUMMARY] vm_swap I/O error during pagein.

From: Ian Mortimer <ian_at_physics.uq.edu.au>
Date: Thu, 13 Jun 2002 09:10:24 +1000

The question was about kernel panics like this:

   vmunix: panic (cpu 0): k_mem_fault: IO error during kernel pagein

There was another panic last weekend and soon after an AdvFS domain
panic on the other partition of the same disk. Analysis of the binary
errorlog also pointed to the disk. The disk was replaced yesterday morning.
So far all is well.

Thanks to redhatcat, alan_at_nabeth... and Derk Tegeler (twice). I'm
including their replies below for the sake of the archives.

---
Ian
---------------------------------------------------------------------
From: "redhatcat"
  You say patched; but not what  patchkit
  1-  If not there already-  Go to Patchkit 6 ( personal good luck with
this )
  2-  if running advfs ; try a defragment on advfs domains
        It nay force the disk issue
From: alan_at_nabeth...
   Start collecting the LBNs of the parts of the disk with
   the problem.  Also get rest of the Request Sense data from
   the event log to see if the ASC/ASCQ offers a clue.  If
   the problem is the same LBN, just replace it by hand.
   If the problem is an area, think about replacing the disk.
From: Derk Tegeler
   Did you get an answer on this one? This looks like a faulty memory chip 
   rather than a disk problem.
   pagein is not the same as a page read or  written from/to disk
From: Derk Tegeler
   You may want to use the exercisers to reproduce faster:
   /usr/field/diskx
   /usr/field/memx
   Do you use Decevent or Compaq Analyse? If not do because it allows you to 
   spot hardware errors much much sooner.

Received on Wed Jun 12 2002 - 23:10:46 NZST