Re: Solaris 10 and 8 or more metadbs cause metadb loss
From: Scott Howard (scott_at_hunterlink.net.au)
Date: 04/02/05
- Next message: Scott Howard: "Re: SCNA Solaris 10"
- Previous message: W: "Re: Xsun support for XWeb"
- In reply to: cade.robinson_at_gmail.com: "Re: Solaris 10 and 8 or more metadbs cause metadb loss"
- Next in thread: Horst Scheuermann: "Re: Solaris 10 and 8 or more metadbs cause metadb loss"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 2 Apr 2005 02:56:36 GMT
cade.robinson@gmail.com wrote:
> I have a bad drive which is obvious with format when it can't type
> it... run explorer.
Are you sure it's not a bad cable/controller? Perhaps it's actually a
hardware array? How big is/was the disk? Is it mirrored? With what
mirroring software? (Remember that some mirroring software needs extra
steps before removing/after replacing a disk). What were the actual
error messages when it failed? (Often important, especially if you
have multiple failures in order to try and work out why)
Sure, you might be able to answer these questions correctly, but you'd be
surprised how many people can't (even if they sound like they can!). The
only way to be sure is to get the data from the machine, and the best
way to do that is to have the Explorer.
> Have bad memory as plainly stated by the machine ... run explorer.
Unless you're running Solaris 10, "bad memory as plainly stated by the
machine" does not mean you've got bad memory. Reported bad memory can
be caused by memory, CPUs, cache, system boards, or even PCI card.
eg, an 'Uncorrectable memory error' message means that the data in
your memory is corrupt (failed ECC checking), but how did that data get
there? It may have been bad when it was written into memory by a CPU, or
by a PCI card (using DVMA) - you need to know how to understand the flow
and the syndrome codes it reports to be able to fully understand what
is broken.
Even if it really it caused by memory, that doesn't mean the correct
action is to replace the memory. We have tools which will analyse logs
from a system and determine exactly what is wrong and what action needs
to be taken - and of course the preference is to run them using Explorer
output.
> The other I have gotten a lot is "Are you on the latest recommended
> cluster?"
Sun engineers should never be giving this response as an answer to a
specific problem. That's not to say that they won't recommend patching
your system if you're a long way behind (which is our recommended best
practice), but they are not allowed to simily give this as a solution
to a problem without researching if this will actually fix the problem.
Scott.
- Next message: Scott Howard: "Re: SCNA Solaris 10"
- Previous message: W: "Re: Xsun support for XWeb"
- In reply to: cade.robinson_at_gmail.com: "Re: Solaris 10 and 8 or more metadbs cause metadb loss"
- Next in thread: Horst Scheuermann: "Re: Solaris 10 and 8 or more metadbs cause metadb loss"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|