Reindl Harald <h.reindl(a)thelounge.net> writes:
Am 17.06.2013 03:44, schrieb lee:
> Reindl Harald <h.reindl(a)thelounge.net> writes:
>> well, most people on Linux are using mdraid to avoid the useless
>> question about "compatible hardware RAID controller" because it
>> is pretty useless to have a RAID with a single-point-of-failure
>> in form the controller
>
> How do these people answer questions like how to connect 10+ SATA and/or
> SAS devices to their computers at the same time, how not to create a
> bottleneck with software raid and how to eliminate every possible single
> point of failure?
why should these people do this with local disks?
Maybe because they want to find to a balance between cost and usefulness
that they are happy with.
such things are done with SAN-storages with typically
*two* conrollers for failover, if you rely on *one*
single controller and nothing else can read your
RAID you should hire someone with knowledge
Yay, sure, I'll send you the bills.
but what the hell has this to do with mcelog.service
It seems to be part of an attempt, directed by invalid assumptions, to
explain how running mcelog all the time might be useful.
since nobody but you is using 10 SAS disks local attached
Why would you assume that I do this?
hint: in such setups *you get* replacement controllers
and disks by SLA's with the manufacturer within hours
Ok, and I send you the bills.
*you said* how can a different machine read my RAID
well, normally this question should be answered
*before* put any data on it
Someone assumed that reading what mcelog might put into a log file was
easily possible even when the hardware is too broken to allow this. I
merely questioned this assumption.
> Are you assuming that most ppl using Linux have fallback systems
with
> independent internet connections and emergency power generators standing
> by that automatically take over seamlessly when something with their
> used system or around it fails?
pff you do not need independent internet connections and power
generators to plug software-raids to whatever computer because
you can plug them with any SATA-to-USB and mdraid will recognize
the array based on UUID's
You're missing the point which was someones assumption that a particular
component must be useless because it can be spf. I was merely asking if
there's also the assumption that ppl go to all the considerable lengths
required to avoid spfs (so that the components they are using aren't
useless).
Don't try to hold me responsible for other ppls logic, flawed or not.
--
"Object-oriented programming languages aren't completely convinced that
you should be allowed to do anything with functions."
http://www.joelonsoftware.com/items/2006/08/01.html