On Fri 14 Jul, 2000, Jay Orr orrjl@stl.nexen.com wrote:
For my griping - everything BUT the motherboard is designed for easy swap-out. Why can't they make the motherboard just as easy? Let's face it, if best-case is 15 minutes on a mission critical system, that's not right. ever taken apart a Sparc or HPUX system? They're designed to be modular so it's just pop-out, pop-in, back in service. A design idea Netapp should follow...
Well, the nearest comparable component in the Suns etc. I think you're talking about would be the back-planes or centre-planes, and they take a very long time to replace indeed.
If you're talking about more NetApp-scale machines like E420R's, rather than Exy00's, I think you'll find the motherboard's are real pains requiring a second torque-adjusted screwdriver (you need one for the CPU modules).
Suns etc. are very variable IMHO - ever tried to put a heatsink back on one of the ASICS or UltraSPARC modules on an Exy00 CPU-board when you catch it accidently sliding it into the chassis? Not easy to do, and often a cause for further FRU shipments.
SGI's are tricky to pull apart in the field too, but they, at least, were intending to lever off the node/module design.
Starfires, IBM S80's and other big boxes, well sometimes they can be serviced online, sometimes they need a day's downtime.
A really rapid mobo swap can be accomplished (in a NetApp, Netra, etc.) by doing a full head-swap if it's *that* important to you.
I have to say that compared to the old 1400's the F540, F630 etc. generation was a substantial improvement in Serviceability, and it looked like the designers had spent some time as field engineers taking notes about how things can be done.
But otherwise, sounds like fun ;)
Aye. I do think points should be awarded for style as well as the final time. Calm, unflapped, neatly executed work by one or two guys may take a hair longer than a biff-bang-bosh job slapped together by three or four adrenaline addicts but I know which I'd rather have serving my data afterward.
-- End of excerpt from Jay Orr
On Fri, 14 Jul 2000, Mark Simmons wrote:
time. Calm, unflapped, neatly executed work by one or two guys may take a hair longer than a biff-bang-bosh job slapped together by three or four adrenaline addicts but I know which I'd rather have serving my data afterward.
Indeed. Though posted in jest, I did not rush the whole affair. I wanted to do this once, and I wanted it to be clean so that our local NetApp engineer wouldn't notice anything sloppy if he were the next person to open the filer.
As hinted at, I unscrewed what I could beforehand, broke the torque or loosened other screws/nuts, had the new motherboard lined up on a table, had static bags lined up for the PCI cards, etc. After halting, it was a matter of just unplugging all the wires, finger unscrewing the hex bolts, etc. The longest segment of time indeed came from the motherboard screws. Installation was the reverse of removal--slide case closed, finger tight all hex bolts, connect all wires, and power up. Final tightening happened during and after boot. Not kosher with the hardware field guide, I'm sure!
Call it cheating or what you will, but being "prepared" does shave off a valuable 5-10 minutes. And, due to lots of swaps in the recent future, practice helps. (Previous times were something like 32 and 25 minutes.) I would not suggest someone try this if they haven't done it a few times before.
Until next time...
The Mathworks, Inc. 508-647-7000 x7792 3 Apple Hill Drive, Natick, MA 01760-2098 508-647-7001 FAX tmerrill@mathworks.com http://www.mathworks.com ---
Maybe we have some new design improvements for Netapp then!
Sounds like we need more edge connectors so that the board slips in and out with the cables etc plugging into an external bus or something?
And for the motherboard reduce the number of screws and what about some sort of easy release system? I'm thinking bicycle wheels here but I'm sure the same applies for motherboard screws!
2 Questions on a slightly different thread tho; what is the life expectancy of a drive, what is the life expectancy of a motherboard?
Steve
-- Stephen J. Wilcox Internet Manager, Opal Telecom http://www.opaltelecom.co.uk/ Tel: 0161 222 2000 Fax: 0161 222 2008
On Fri, 14 Jul 2000, Todd C. Merrill wrote:
On Fri, 14 Jul 2000, Mark Simmons wrote:
time. Calm, unflapped, neatly executed work by one or two guys may take a hair longer than a biff-bang-bosh job slapped together by three or four adrenaline addicts but I know which I'd rather have serving my data afterward.
Indeed. Though posted in jest, I did not rush the whole affair. I wanted to do this once, and I wanted it to be clean so that our local NetApp engineer wouldn't notice anything sloppy if he were the next person to open the filer.
As hinted at, I unscrewed what I could beforehand, broke the torque or loosened other screws/nuts, had the new motherboard lined up on a table, had static bags lined up for the PCI cards, etc. After halting, it was a matter of just unplugging all the wires, finger unscrewing the hex bolts, etc. The longest segment of time indeed came from the motherboard screws. Installation was the reverse of removal--slide case closed, finger tight all hex bolts, connect all wires, and power up. Final tightening happened during and after boot. Not kosher with the hardware field guide, I'm sure!
Call it cheating or what you will, but being "prepared" does shave off a valuable 5-10 minutes. And, due to lots of swaps in the recent future, practice helps. (Previous times were something like 32 and 25 minutes.) I would not suggest someone try this if they haven't done it a few times before.
Until next time...
The Mathworks, Inc. 508-647-7000 x7792 3 Apple Hill Drive, Natick, MA 01760-2098 508-647-7001 FAX tmerrill@mathworks.com http://www.mathworks.com
For the "Design Improvements" I would like to nominate the following:
A panel on the front, ala Sun's e series, with a removable clear panel that you can put in labels for names and IP's.
We have over 40 of these machines and wind up having to put stick labels on the front so we can tell which machine is which. If/when we rename or assign a new IP you have to scrap off the old label which scratches the front or leaves a sticky residue on the bezel. This way you could jus pop of the clear plate and stick a new piece of paper in there with the new info!
-----Original Message----- From: owner-toasters@mathworks.com [mailto:owner-toasters@mathworks.com]On Behalf Of Stephen J. Wilcox Sent: Friday, July 14, 2000 9:51 AM To: Todd C. Merrill Cc: toasters@mathworks.com Subject: RE: filer Olympics
Maybe we have some new design improvements for Netapp then!
Sounds like we need more edge connectors so that the board slips in and out with the cables etc plugging into an external bus or something?
And for the motherboard reduce the number of screws and what about some sort of easy release system? I'm thinking bicycle wheels here but I'm sure the same applies for motherboard screws!
2 Questions on a slightly different thread tho; what is the life expectancy of a drive, what is the life expectancy of a motherboard?
Steve
-- Stephen J. Wilcox Internet Manager, Opal Telecom http://www.opaltelecom.co.uk/ Tel: 0161 222 2000 Fax: 0161 222 2008
On Fri, 14 Jul 2000, Todd C. Merrill wrote:
On Fri, 14 Jul 2000, Mark Simmons wrote:
time. Calm, unflapped, neatly executed work by one or two guys may take a hair longer than a biff-bang-bosh job slapped
together by three
or four adrenaline addicts but I know which I'd rather
have serving my
data afterward.
Indeed. Though posted in jest, I did not rush the whole affair. I wanted to do this once, and I wanted it to be
clean so that
our local NetApp engineer wouldn't notice anything sloppy if he were the next person to open the filer.
As hinted at, I unscrewed what I could beforehand, broke the torque or loosened other screws/nuts, had the new
motherboard lined up
on a table, had static bags lined up for the PCI cards, etc. After halting, it was a matter of just unplugging all the wires, finger unscrewing the hex bolts, etc. The longest segment of time
indeed came
from the motherboard screws. Installation was the reverse of removal--slide case closed, finger tight all hex bolts, connect all wires, and power up. Final tightening happened during and
after boot.
Not kosher with the hardware field guide, I'm sure!
Call it cheating or what you will, but being "prepared" does shave off a valuable 5-10 minutes. And, due to lots of swaps in the recent future, practice helps. (Previous times were
something like 32
and 25 minutes.) I would not suggest someone try this if
they haven't
done it a few times before.
Until next time...
The Mathworks, Inc. 508-647-7000 x7792 3 Apple Hill Drive, Natick, MA 01760-2098 508-647-7001 FAX tmerrill@mathworks.com
We have over 40 of these machines and wind up having to put stick labels on the front so we can tell which machine is which. If/when we rename or assign a new IP you have to scrap off the old label which scratches the front or leaves a sticky residue on the bezel. This way you could jus pop of the clear plate and stick a new piece of paper in there with the new info!
Try "citrus degreaser" for helping remove labels. Use a rag damp with the degreaser to apply directly to the paper, it will soak in and unstick the adheasive. Labels should peal off much easier. I've used this trick a few times, though, not on a Netapp (but have on PC cases, laptops, etc), and it seems fairly safe. Its non-toxic and smells good too. Citrus degreaser can be purchased at most bicycle stores.
-
Dan Brown -
Lance R. Mehle -
mds＠gbnet.net -
Stephen J. Wilcox -
Todd C. Merrill