WD20EARS just added on the official HCL - mistake and issues

adamwilt wrote:
Has anyone tried installing the SSH add-on for an Intel-based ReadyNAS and fiddled with the logging setup and other runtime parameters, as described in the WD article, to reduce the "idle time" hits on the disks? Did it make any difference in how quickly the LCC increases?

I answered my own question: the disk-sync disabling is already in place in the syslog.conf setup, and a --MARK-- every 20 minutes couldn't account for the LCC increases folks see.

So I took CommanderQ's ISO image for a WDIDLE3 utility (at viewtopic.php?f=24&t=39737&start=95) and set differing idle timeouts on WD20EARS disks in my new Ultra 6. Over a 15-minute period (with about a 20% duty cycle of accessing vs leaving the NAS idle), here's what I saw in terms of LCC increases:

Idle setting, seconds: LCC increase, count/15 min
8.0 (default): 31
10.0: 23
12.5: 15
30.0: 0
60.0: 0

WDIDLE3 version 1.05 only allows settings from 8.0-12.7 seconds and from 30.0 - 300.0 seconds; just looking at the trends I wonder if 15 or 20 seconds would have been fine... but 30 works, so I wound up setting all my disks to 30.0, and they've been happy ever since. And if by some chance they do get left alone, they will unload and save power.

An odd thing: we've been using WD20EADS (prior generation) drives in our two existing ReadyNAS Pros for the past year, and in a couple of our Macs. The WD20EADS *also* has an idle time of 8.0 seconds according to WDIDLE3, yet the LCC on these drives hasn't been climbing excessively; it's roughly in line with the start/stop count. Why aren't the EADS drives showing the same LCC behavior? The EARS has 4k sectors and twice the cache of the EADS, but should those alone result in such a radically different load cycle count?