[nSLUG] Multiply-claimed blocks?
draker at gmail.com
Wed Aug 31 14:34:33 ADT 2011
I did not say 'none' because I was not in charge of the monitoring.
People who were sometimes had to hunt around for the sysadmin would
admit to making a change without having properly logged it. So I can't
be certain how many changes might never have been accounted for.
However sysadmin changes are typically in text config files, or a set
of binary files in an upgraded package. I never heard of an
unexplained single binary file checksum failure.
The fact that the prof's computer was a shared resource doesn't seem relevant.
It's my OPINION that there must have been something wrong with that
computer (not just cosmic rays). I think that if a bit flip on disk
once a month was "normal", our computers would be crashing a whole lot
On 31 August 2011 14:26, Dop Ganger <nslug at fop.ns.ca> wrote:
> On Wed, 31 Aug 2011, Daniel Morrison wrote:
>> On 31 August 2011 14:12, Dop Ganger <nslug at fop.ns.ca> wrote:
>>> I also vaguely recall hearing about a prof who ran a daily md5 of all the
>>> files in his ~, and regularly found checksum errors.
>> Sounds to me like that prof had a bad hard drive (or maybe his
>> computer was at a high altitude?)
> This was a shared computer. He did it because everyone else believe the
> system was perfect as long as there was no error, and he believed there wree
> intermittent errors coming through. I think he was seeing an error every
> month or so.
>> At my old job, for many years, all system files on some dozens of
>> machines were checksummed nightly, and any discrepancy had to be
>> explained. There were no (or so few as to be negligible) occurrences
>> of spontaneous changes.
> Few != none.
> I suspect this is a problem that won't be seen in a system that's protected
> by ECC or similar all the way through, but the system I'm talking about was
> a shared system for departmental use - I think an Ultrasparc of some
> Cheers... Dop.
> nSLUG mailing list
> nSLUG at nslug.ns.ca
More information about the nSLUG