Re: [Sks-devel] more database corruption

[Dan Egli]

On Sun, 2 Nov 2003, Dan Egli wrote:

> On Sun, 2 Nov 2003, Dan Egli wrote:
> 
> > On Sun, 2 Nov 2003, Dan Egli wrote:
> > 
> > > I'm running a badblock check on the partition that the database resides
> > > on right now. If that comes up empty I'll run a memory test. Beyond that
> > > I don't know what hardware could be bad.
> > > 
> > 
> > 
> > Well the badblock scan only found one bad block, and it's the 2nd to last 
> > block on the partition so I doubt that bad blocks on the drive is to 
> > blame. I'm gona take the server down for a few hours to run some memory 
> > tests. Perhaps I've got some bad ram or something.
> 
> The quick ram test came up 100% ok. I'm going to leave it for an extended 
> test, but beyond that, I don't know.
> 
> I'm also going to run a hard badblock scan, repeating till it finds no new 
> blocks. Perhaps it's an intermitant failure.


extended ram test shows no errors, extended hdd test only found the one 
bad block at the end of the drive (anyone know how I set that block as bad 
on an ext3 w/o reformatting the partition?), so I'm suspecting the backup 
routine. The database has died twice during that routine now. I'm going to 
disable the backup routine tonight and see if the database remains intact. 
If so then we need to figure out why the routine is killing the database.

--- Dan