Re: Serious Crash last Friday

On Wed, 10 Jul 2002, Andrew Sullivan wrote:

> On Wed, Jul 10, 2002 at 05:19:47PM +0200, Henrik Steffen wrote:
> > Hi,
> >
> > thanks for the information...
> >
> > the badblocks read-only test did not report any problems,
> > do you think i should run the "read-write" test, too?
>
> Well, if you do it'll destoy the data, so although it's the only way
> to be sure, I wouldn't unless absolutely pushed to do so.  A
> read-write badblocks test on a big partition can take many hours.

This isn't entirely true.  According to bad blocks' man page:

-n     Use  non-destructive  read-write  mode.  By default
       only a  non-destructive  read-only  test  is  done.
       This  option  must  not  be  combined  with  the -w
       option, as they are mutually exclusive.

So, with the -n switch, badblocks will save a sector, do a write / read
test, then restore the sector.

Note that this is pretty slow, as I've tested it before.

> > tonight I will have the memory checked by memtest86 ...
>
> Yes, that seems a good idea.  Brand new hardware doesn't guarantee
> anything, particularly when memory is so fast these days (I've had
> DIMMs fail a couple of months after they were new).

Also, another REALLY good test for bad memory is to build postgresql from
source a couple dozen times, especially with a -j switch set to about 6 or
so.