D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] Q for the server builders

 


Thanks for the comments - James & Dave.

I'd probably rather have 2 drives myself, but clients on a budget, etc... So I'm going to weigh it all up when I speak with them next.

With regard to capacity - they don't need 1TB - I anticipate them at about the 300GB level in 3-4 years time when I'll be looking to replace the servers, however I've been building boxes with 2x the required disk capacity for a long time now - I keep a local archive partition updated nightly using rsync, etc. which gives me a few days worth of accidental deletion protection. Often a lot easier than restoring from another server or tape (in the days when I used tape!) Price wise - there isn't a lot in it these days - Aria has WDC 1TB drives for £55+VAT today... (One offs though)

I've also installed quite a few of the WDC Green 1TB drives. No issues so-far and they run nice and cool. Power consumption is another issue here - the mobos are likely to be Intel Atom ones. Not as lean as VIA boards but a lot more grunt.

Remote hands are essentially free if it's an "easy" job during office hours - which it would be...

Dave wrote:

Personally, I quite like the thought of getting an impending doom email,
"member failed in array" - rather than the server just not responding to
ping, or other Nagios etc' failure.  Further, from my experience, some
drives appear fine - but tend to 'fail' when the server is rebooted,
therefore having RAID gives that extra security blanket of both servers
failing in the same timeframe.

Personally, I'd rather get and respond to the warnings generated by the SMART stuff before the "oh sh*t" email from mdadm!!! I've not had a drive die in recent years that hasn't issued SMART warnings first - actually, I've not had a drive fail catastrophically since the days of the 80GB Maxtor and IDM Deathstar drives, however...

And recent mdadm and Linux raid drivers have a lot of support for active disk checking these days too.

As for rebooting a server - who does that anymore? ;-)

  % uptime
   12:30:34 up 1242 days, 23:33,  1 user,  load average: 0.00, 0.00, 0.00

Actually I do sort of dread rebooting that one - hopefully it'll be retired soon...

The HA stuff will take care of a server dieing - or ought to if it's setup correctly.

Ah well - time to get back to the garden and ponder all this over..

Cheers,

Gordon
-- 
The Mailing List for the Devon & Cornwall LUG
http://mailman.dclug.org.uk/listinfo/list
FAQ: http://www.dcglug.org.uk/linux_adm/list-faq.html