hi guys, just to follow up on that load cycle thing ive seen crop up in a few posts, i had a few easy bite-size questions please
is it realistically in issue?
(eg my drobos are on almost 24/7 and using WD GP’s eads drives)
i have nightly backups for offsite which trigger drive usage during the night for around 1 hour, and sometimes leave long synch jobs running on them, (but i guess theres a fair amount of IDLE time too).
do you mean that if i dont use the drobo (accessing/writing data) for say 10mins, and the drives spin down - is the spinning up again what you called CYCLing? if so i might hear that happen, (and then wait about 5 seconds per drive) for the spin backs, and then i can access files etc) probably a few times a day.
Is this like that MTBF rate, which works out to be say 5-10 years or something, whereas i’d probably have swapped out these existing 1TB drives by then, and its sort of a false panic?
doc chris recommended me the EADS gp’s - where are you hiding LOL - (joke, so far everhthing works a treat my friend)
as far as i know it doesnt affect drobos (or at least not severely enough to worry about)
the problem is that the heads are designed to be taken off the disks when its idle… the disks wait 8 seconds - then decide its idle. thats fine for windows and osx but under linux apparently it quite often doe data transfers in bursts so the disks load and unload about every 10 seconds which wears them out. drobos (the regular ones - im not sure about FS) run VXWorks (i believe) and i dont know if that triggers the loads and unloads or not…
the point of unloading the heads is that it reduces wear and reduces power consumption, so its a good idea, but it shoudlnt be toggling it ever few seconds
the heads are rated for 10,000 Or maybe 100,000 load cycles. usually that should be more than 10 years - ie. more than the drives - but if you are only reading and writing every 10 seconds like linux tends to then you can exceed that rating only 6-9 months which causes your disks to die
the EADS and EARS both have this issue ( i think some firmwares have it set to 30 rather htan 8 seconds which complete solves the issue)
you can use a utility called WDIDLE to manually set the timeout - i have set it to 5 minutes on all my drives - so if the drive is idle for 5 minutes it unloads the heads[hr]
and i have a very good sense of humour
is there a procedure for using that WDIDLE tool?
eg can i simply use it while the drives are in a drobo, as long as i boot up with a boot disk, or would each drive need to be taken out 1 at a time, and put in a separate sata usb caddy case?
(if the only way is to open up my oldmachine and try and find a sata connection to plug it to, then i’ll probably give it a miss)
it has to be a direct sata connection to a DOS environment (which is why the USB caddy wont work - USB wouldn’t function)
its a real PITA and who knows if it is even an issue on a drobo
power down your drobo… connect a drive to your PC (via USB will be fine) and use one of the free tools to see the drives SMART information and see what its load/unload cycle count is - if its less than several thousand, then you have nothing to worry about
people are complaining when they are seeing 10,000 loads in a week (since that will wear it out).
then put the drive back into drobo and power it back on
(since you were only reading its smart information it wont have changed anything)
