[AccessD] A real puzzler

[John W. Colby]

>An aside Re drives - if you will use Seagate - check the 4TB 7200.14 (ex barracuda) power usage, or are you into 5 year warrantee ones

I don't.  Have you read this?

https://www.backblaze.com/blog/hard-drive-reliability-update-september-2014/

I wouldn't touch Seagate with a ten foot pole, even before reading this.

Re second system - don't you have an almost complete (used) set of the bits
needed.

Well... I used to, but not really.  I would need a power supply and server chassis.  I have the motherboard and an old (but still functional) Areca Raid Controller.  I would need several new hard drives.

So chassis and power supply would be about $600-$800.  A hand full of 3tb drives would be another $600 or so.

Then it depends on whether I want to try splitting the CPU and memory in two and using 1/2 on the new system. I have two physical 8 core AMD processors, so I could move one into the new system.  The memory sticks are 8 GB IIRC so I could steal at least 32 gigs from the existing system.

And would doing this transfer the problem to the new system?


>>Re ECC - Should deal with actual memory cell problems, but I suspect may not deal with insufficient power or refresh vs read cycles

Yea, I never really played with the memory stuff.  I really just put it in, turned it on and left it running.  What I can say is that I always purchased the exact same manufacturer and part number, though as it turns out even then they do change slightly over time.  And the system worked reliably for several years.  It is entirely possible that it was when I added more memory that the problem started.  At first the reboots were very far apart, once a month or even less, so I was never able to really pin down / remember "what I had done last".

And why has the frequency of the reboots gradually increased, from once a month to several a day?

The whole thing is just strange.

John W. Colby

On 8/8/2015 8:34 AM, James Button wrote:
> >From the concentration of bmmrs listed and the changes already done, I'd
> definitely look at power management as a probable trigger.
>
> Re PSU not faulting under 'No load' - unusual I'd agree, and you've
>   1) changed that, and
>   2) obviously have a quality one installed
>
> It's in a proper place, presumably with lots of  'friends'  - and you've had the
> problem in different locations,
> so that would rule out external causes.
>
> Re ECC - Should deal with actual memory cell problems, but I suspect may not
> deal with insufficient power or refresh vs read cycles
>
> An aside Re drives - if you will use Seagate - check the 4TB 7200.14 (ex
> barracuda) power usage, or are you into 5 year warrantee ones
>
> Re second system - don't you have an almost complete (used) set of the bits
> needed.
>
> Re windows activity - 'Idle-time' indexing of all files on a partition,
> Idle-time restructure of MFT - maybe event recording might show something, but
> I'd still be looking at 'power management'
>
> BUT -
> Whatever - I'm still guessing from experiences with older hardware, from (way
> way) back when I was told I was now the 'Admin' person for the corporate server
> -  2 x 2GB and 4 x 1GB PATA drives with backup to caddy'd hard drives and CD.
>
> JimB
>