RAID boards and standby?

V

Valnar

Supreme [H]ardness
Joined
Apr 3, 2001
Messages
4,570
I have a Highpoint 2320 RAID card running on my Windows 2003 Server. I have no problems with it and everything works well, except one. It does not support a graceful return from Standby mode. I have to keep my server running 24/7. That may or may not be a ding against Highpoint though - it's my understand most RAID cards have this issue.

So my question is.... short of returning to my built-in Intel ICH onboard controller, is there any decent aftermarket PCIe RAID board out there that supports standby, where all hard drives can spin-up reliably and work?
 
It's an issue with PCIe power state and how the card handles it. Likely the HighPoint does not reissue MOTOR_START when returning from S1/S3 to D0. These are not designed for S1/S3 operation typically, so it's not unexpected. I'm not aware of any cards that will attempt a MOTOR_START from S1->D0 or S3->D0.
You can poke around the card settings and see if maybe there's an option to do a CMD_READY check on power state transition, but otherwise, it'd be a firmware change.
 
Do you know if a PCI-X board would behave better? My Supermicro board has those too.
 
Nope. PCIe just expanded on PCI power states. Still the same power states and the same behavior. RAID controllers aren't meant to be suspended like that ever, especially those with BBUs. In S1/S3, the BBU is being discharged to maintain cache. This is a bad thing.
 
Thanks... maybe I'll just upgrade to one that handles spindown.
 
Valnar said:
Thanks... maybe I'll just upgrade to one that handles spindown.
Click to expand...

None of the controllers on the market that I'm aware of (and that's a lot of controllers) will do a spindown and spin up from S1/S3. They're just not designed for that - they're designed for systems that are always on. The best you can hope for is a controller that does a CMD_READY poll and attempts MOTOR_START on the failure. I don't know of any that do that behavior, partly for lack of testing. Most controllers will flag a drive as failed if it spins down of it's own accord, rather than attempt to restart. Flag all the drives failed, and disaster results.
 
According to Areca their card do support standby mode:
(And seemingly any driver that passes WHQL for Server2003 needs to support it.)

http://faq.areca.com.tw/index.php?o...items&cid=7:Windows&id=140:Q10110603&Itemid=1

If this is the case, as I'll have to test later on, I'm going to keep my server in standby mode. Currently I just have some power saving CPU and disk features turned on. (lower clock+voltage at idle, spindown after ~15min)
 
pissboy said:
According to Areca their card do support standby mode:
(And seemingly any driver that passes WHQL for Server2003 needs to support it.)

http://faq.areca.com.tw/index.php?o...items&cid=7:Windows&id=140:Q10110603&Itemid=1

If this is the case, as I'll have to test later on, I'm going to keep my server in standby mode. Currently I just have some power saving CPU and disk features turned on. (lower clock+voltage at idle, spindown after ~15min)
Click to expand...

Even that would be an improvement. My Highpoint does not support spindown.
 
Yeah, but there's a difference between supporting S1/S3 and actually working with S1/S3. Every RAID card on the market support S1/S3 power states, that's just a standard requirement these days. The problem is that when returning from S1/S3, the drives are spun down. There's no requirement that the cards work right when doing S1/S3->D0, only that the card itself come back. I can verify that that works just fine, when you don't spin the drives down. I can't verify that Areca will restart the drives after spindown and S1/S3 transitions. (And no, just spinning drives down does NOT count. It is NOT the same.)
 
You must log in or register to reply here.
Back
Top