Adaptec's Series 5 Unified Serial Controllers Dissected

[rozar]

I have done some thermal testing recently with the 5 series Adaptec. I tried to create a "worst environment" to start. I used an Intel mid tower chassis (5299) with just the rear fan. I installed 6 drives and the controller below with no active cooling at all, just the 120mm rear fan. With this setup I was able to get the controller to 201 degrees using IOmeter, at which time it did give a warning in the Adaptec Storage Manager software. No data loss, just a warning. I was originally curious as to just hot this thing would run and still be "normal" before it gave a warning. It appears that less than 200 degrees is "that" number. (Although I would not want a card of mine running that hot) Next I switched to a drive housing that has a fan on the back of it and also put a 120mm fan in the front of the chassis where there is already a mounting location in the blue plastic insert that goes in the front of the chassis under the drives. This did not provide much air flow to the heat sink of the controller but it did move the air. (Keep in mind that the 120mm fan was not spinning very fast) This small amount of air was able to bring the controller temp down to 126 degrees using the same IOmeter test, which is not bad at all.

So the long story short of this is that only a small amount of chassis air movement is needed to cool the 5 series cards. But it must have at least some active cooling.

 

[Turas]

Interesting, on their site I think they have it rated to 60c which is why I was so worried. That is very good to know. Thanks for the information.

 

[Turas]

I do have Ubuntu up and running on the ARECA card right now and it runs good. I prefer to disable my onboard controllers all together and hook up even my CD?DVD drives to the SAS controller. With the Adaptec they handle everything just fine. It seems that the ARECA one does not recognize DVD drives. I can't find anything about it in the documentation but I figured since it did SAS/SATA that it would handle those too. I am still on the fence about keeping this or going back to the Adaptec one. I still get corruption running Windows. It seems to occur after I install my NOD32 AV and then reboot. I thought maybe it was NOD then but when I put the adaptec in I can load Windows and my AV and have no problems. So it seems the ARECA cards may not be as stable as the Adaptec ones. At least from my point of view. What to do?

 

[wilfy]

I bought an Adaptec 5805 on the strength of the article from span.com. It's an expensive card and I bought it to replace a Highpoint RocketRAID 2320, which seemed to 'disappear' after a week or so's use without a reboot.

Firstly, the card runs hot. It started alarming after a few minutes running, so I made sure that there were case fans blowing on the card and the card ceased alarming.

Perhaps I am doing something wrong, or perhaps my card is faulty, however if I build an array, decide I got it wrong and delete it, then none of the drives from the deleted array are available for creation again (?). Experimenting showed that if you pull them all out while it is running (!), you can do what you want 8-{

FX: Not impressed!

Anyway, on with the build. Fortunately Linux is up and running on a mirrored pair using mdadm, and I want to restore from my archives on a backup drive to the new array. I have seven drives in RAID 5 and one global hot spare.

I want to build the driver from source so first I have to extract it from the rpm on the driver CD, it's v1.1-5[2458]. It would be nice if there was a nice bz2/gzip tarball for those not using RedHat or derivatives. I backup the kernel bundled driver and install the source in drivers/scsi/aacraid. I select the aacraid driver in make menuconfig, and the kernel builds and installs without a hitch.

Great!

I boot up, partition up the drive, make the partitions into ext3fs and start extracting the files. A hundred Gb or so later, the array goes read only and the unarchive is aborted.

When the machine is rebooted, only four drives out of the seven seem to be in the array, the rest are greyed out.

I check out the firmware - it's on v5.2-0 build 16116. I can see that the card is already running the latest firmware. I download the latest driver source from Adaptec and I am rebuilding the array right now and have built the kernel with v1.1-5[2459].

It's not looking good.