[OpenIndiana-discuss] Are there any issues re ZFS on 2TB disks (from WD??) with large blocksizes?
gdriggs at gmail.com
Wed Feb 9 12:22:59 UTC 2011
On Feb 9, 2011, at 3:06 AM, Matt Connolly wrote:
> I stil get these symptoms with OI-148.
>> 2. They spin down heads every 8 seconds.
> This particular problem can be overcome by using the WD3IDLE command to defeat this behaviour.
>> 3. I have seen a few bad sector reports and bad heads on those drives.
> I haven't seen any read/write errors in `zpool status` ever. (nearly one year old).
> I'm not going to get another WD * EARS drive.
This and another thread about ECC vs non-ECC memory got me thinking... I don't know how this applies to ZFS but under Windows XP there's an "advanced format align utility" that WD has for aligning 512 byte writes on 4k boundaries.
But what really stands out is that you're using a desktop drive & expecting enterprise features from it. The white paper below was written shortly before 4k sector drives but it's still relevant. One thing it doesn't mention that I often hear from junior sysadmins is some clarification about OEM server drives vs off the shelf desktop drives -- a conversation usually started over pricing differences.
This is becoming less of an issue in the enterprise because fewer servers use SATA drives these days but if you're building your own array using SATA drives or some mix of SAS and SATA in the same chassis then it's something to keep in mind. OEMs often have their own disk controller firmware tweaks and burn-in testing procedures that are applied before shipping. Without confusing the issue by talking more about OEM vs standard enterprise drives in this message, please consider the details in the white paper before deploying drives for home or office in a system that you're expecting to use for long term retention of data.
More information about the OpenIndiana-discuss