RocketRAID 2314 alarms when external enclosure is turned off

[flatfoot]

Hi there,

I just installed my brand new RocketRAID 2314 in my Mac Pro (SL 10.6.2).
Everything seems to be fine except for two things:

1. There are no eject-signs next to the eternal drives' icons in Finder windows. I can, however, eject them by highlighting them and hitting cmd-e.

2. And that's the actual problem: When I switch off the external enclosure after ejecting the drives properly, the card's beeping alarm goes off.

Isn't the RR2314 even supposed to be hot-swappable? Then it should by all means let me switch off the enclosure without starting to whine, shouldn't it?
Or is it a firmware issue? I already tried to flash the card with the latest firmware using HighPoint's linux-flasher and a Knoppix CD, but the flasher couldn't find the card (which _is_ properly installed).

Any help would be highly appreciated!

[cvelezv]

You have to enter the High Point web application (using safari) and eject the drive(s) from there, this must be done after ejecting the drives from Finder. To access the web app use: https://localhost:7402, the default username is RAID and the initial password id hpt. After login go to device management and select "unplug"

[flatfoot]

Thanks a lot – that should solve my problem.

Unfortunately, the WebGUI didn't load and when I tried to start hptsvr manually, it said "no drivers detected. Abort!"
After crawling threw the HighPoint-pages I finally found the latest versions of the drivers and the WebGUI – now it works.
Shouldn't SL update at least the drivers automatically? I mean, they are incorporated in the installation...

[KonaBlend]

Originally Posted by flatfoot

Thanks a lot – that should solve my problem.

Unfortunately, the WebGUI didn't load and when I tried to start hptsvr manually, it said "no drivers detected. Abort!"
After crawling threw the HighPoint-pages I finally found the latest versions of the drivers and the WebGUI – now it works.
Shouldn't SL update at least the drivers automatically? I mean, they are incorporated in the installation...

It's possible they want to get some solid feedback on the new drivers before availability via Apple's update.

That alarm is really annoying and while other HighPoint adapters can disable via physical jumpers, the 2314 can at least (thankfully) disable it via web software.

On a related note, I've found it's important to run HighPointRR 1.1.1 or newer. IIRC Snow Leopard bundles HighPointRR 1.0.0 which seems to kernel-panic when in 32-bit kernel mode (64-bit didn't have this issue).

[flatfoot]

Thanks for the info.
After installing the latest drivers and WebGUI, I also updated the card's firmware (from 2.3 to 2.5) via the GUI and I'm absolutely satisfied by its throughput. Just making a new TimeMachine backup and I get between 100MB and 140MB sustained read/write from my internal RAID0 to the external one.

My next upgrade is to add another two disks to my internal 2-disk RAID0.

EDIT: Wait... So it's ok to turn off the alarm? Will my external drives be recognized again if I turn them off and on again some time later?

[KonaBlend]

Originally Posted by flatfoot

EDIT: Wait... So it's ok to turn off the alarm? Will my external drives be recognized again if I turn them off and on again some time later?

Yes, it's ok to turn off the alarm, and hot unplug/plug cycle works for me. I haven't tried a the sequence of unmount/poweroff -> poweron/mount yet but I suspect that will work just fine. I'm running raid 1+0 and have been able to pull a drive, and put it back in, and rebuild the mirror. Apple's software raid has a few quirks but is usable.

On a throughput note, I did test putting 4 x 1TB 7200rpm drives (seagate) direct (ie: no port multiplier) and the throughput is excellent. Streaming a large file, it starts off around 450MB/s with raid 0, empty HFS+, and as the filesystem starts to consume tracks inwardly, 240MB/s. Readback was near identical in performance. raid 1+0 did the expected thing and chopped throughput in half, for both writes and reads. The streaming performance of reads on a raid 1+0 did not benefit from any smart round-robin. Maybe it helps for IO ops but I haven't measured that yet.

Technical tidbit - bypassing the filesystem layer and just doing raw block reads on the disks, I was able to sustain 480MB/s at outer tracks, and iirc ~290MB/s at inner tracks (the sum total of all 4 disks concurrently).

Avg to read a single entire raw disk was 109MB/s, roughly 2.5 hours -- and yes, the card was able to read all 4 disks concurrently, same throughput for each disk. (MacPro3,1 PCIe x4 slot).

All in all, very happy with the performance of this card and it's a fine choice for JBOD and to do sw-raid or similar.

...and that 450MB/s is so blazingly fast, almost want to keep it at raid 0