I/O Errors with Perc H310

Question

Hoping someone here can help me out. I have been banging my head trying to figure this out and to no avail.

We have a new t420 with a Perc 310 running 2008 R2. I ordered the computer from Dell with a single 2.5" 7200RPM SATA drive (500gb) and a 16 hot swappable front panel. We wanted to add more drives to the system mostly SSDs that the customer owns. There are a total of 4 Samsung 840 SSDs. To install I found a seller of some second hand caddies as Dell would not sell me any caddies. I also install two regular hard drives (2.5) into the system one as a mirror (raid 1) for the drive that I ordered the server with and the other as a backup for the mirror OS drives.

Problems started when I attempted to setup windows backup to backup the OS mirror to the other hard drive in the bay. It failed saying it had an I/O error. I thought at first that perhaps I had a bad hard drive out of the box and setup the backup to go to an external hard drive which has been working fine while I work on a replacement drive. That said, I dont think I actually have a bad drive as yesterday we started having trouble with Quickbooks Database Server Manager crashing on the server and throwing I/O errors in the event viewer. These quickbooks files were stored on one of the 4 SSDs I installed. I did a good deal of testing to get QUickbooks to be stable but was only successful when I moved it to the actually C drive of the server and off the SSDs. Also I have had trouble with CrashPlan which is a backup software we were using for cloud backup. It crashes as well when it attempts to backup the SSDs. This software is normally very stable so again odd. Unfortuntely, it is not logging errors that I can point to I/O.

All that said, I spoke with Dell Pro Support and the only thing they can point to is that the drives I have installed after the fact are not "certified" and are therefore showing up as explanation points in the RAID Manager OpenManage. Basically, I am stuck. Not sure if anyone else has had issues like this before or can give me any direction on how to troubleshoot. I really would like them to be able to use these SSDs but at this point I am unsure what else to do. That said, I do not believe it to be just the SSDs as I am also having trouble with the conventional hard drive that was supposed to be a backup to the RAID. It is a WD Black 500GB.

Thanks in advance for your help.

theflash1932 · Accepted Answer

Not sure that anyone will be able to help you get non-certified drives to work correctly. There have been several reported issues with off-the-shelf/consumer SSD's on Dell controllers. The firmware simply doesn't have all the answers to the controller requests ... in some cases, the drives don't even show up on the controller, in some cases, they show up but can't be configured, in some cases, they randomly fall offline, and in every case I've seen, they show erroneous status LED's (showing offline when they are online - a sign they cannot effectively communicate with the controller). You may consider trying to leave the drives unconfigured (no RAID) and letting Windows mirror them ... maybe the problem won't be present in non-RAID mode.

The H310 is a very low-end controller, so expectations should be adjusted accordingly (performance, flexibility, reliability, compatibility, etc.).

What problems exactly are you having with your conventional HDD?

GammaTech · Answer

Any advise from anyone? Really desperate to get this fixed.

GammaTech · Answer

Thank you for the response. As I have progressed I have made some progress. Basically my setup is as follows:

500gb 7200RPM - Seagate

500gb 7200RPM - WD Black

Above drives in Raid 1

4 512GB Samsung 840 SSDs configured as straight drives no raid. Used as file shares

1TB WD Blue 5400RPM - supposed to hold backups of the OS

Windows backup will fail consistently if I attempt to backup the C to any of the drives in connected to the raid controller. I have tried all 4 SSDs as a destination and the 1TB blue. I do not have high speed expectations or am I complaining about speed it is simply that I am unable to get things to behave in a stable matter. If I run windows backup of the C to an external drive I have no problem. It consistently works. This is just one issue. The issue is extended by similar issues with crashplan (another backup software) and Quickbooks. Both of which are complaining of strange I/O errors such as "disk full" when the disk has over 400GB free or just plain crashing.

As additional info from yesterday, today windows search indexer began to have some problems and I ran a bench mark on the SSDs. They perform terribly in the random write IOP test and only get 1400-1500 IOPs when they should be getting upwards of 30,000 in the worst case and up to 70,000.

Any ideas? Could i have a bad raid card? I understand that having a perc h310 I should not expect much but reliability I assume is expected regardless of the card.

Dark-Yak · Answer

Hi all, just sharing my feedback on DELL PERC and Samsung SSD...

I also use the PERC H310 (but the "mini" version for blade server). I have 3 blade servers with the same firmware version 20.12.0-0004.
Each server has 2 x Samsung 840 Pro (Raid 1), each Samsung drive have the latest firmware DXM05B0Q

For unkown reasons, the CMC M1000e enclosure displays errors on 2 blades servers whereas it is working fine with the first server ?!?
On the 2 servers (where errors are reported), the Raid1 volume is working fine...

I have also 2 R610 servers with Perc H700 with 4 x Samsung SSD 840 and they all work fine...

I didn't figure out what makes the SSD work or fail with the different configurations...

François

GammaTech · Answer

Thanks- as a note to future we ended up replacing the raid card with a new h710. 2 days in production and no issues so far. Something we also did different is we configured the drives that were just physical drives with no raid to individual raid 0 virtual drives. Hope that helps someone else. If it starts acting up again ill post again.

Dark-Yak · Answer

Hi all !

I have found some differences between my 8 SSDs, i have the following P/N (product number, do not confuse with the serial number S/N, these information is written on the back of the SSD):

- 3 x MZ7PD512HAGM

- 5 x MZ7PD512HCGM

(notice the difference on the letter A and C at the end of )

The 3 MZ7PD512HAGM are working fine in the my blades (H310) while the others SSD generate alarms on the H310 (i have also tested them on the PERC 6/i and on a SAN MD3620i. By the way, the good SSDs MZ7PD512HAGM were not rejected by the SAN, no alarm, but were not recognized as a compatible drive for the SAN)

Dark-Yak · Answer

Hello again !

So to keep this thead updated, as i have 5 working SSD (MZ7PD512HCGM) but with alarms on my blades, i buy a new R720XD with a PERC 710P (mini version) and... tada, the same alarms ! the Perc displays an alarm (the SSD LED blinks orange).

As soon as i inserted a SSD with P/N MZ7PD512HAGM (notice the A instead of C), the SSD shows green LED...

I don't know il there is something related with the mini version of the Perc (H310 for the blades or H710 for the R720XD) because GammaTech has successfully made his SSD works the new Perc H710.

@GammaTech : Does your SSD blink Orange even if they are working (like my SSD) ?

Bye, bye,
François

GammaTech · Answer

No- but strangely the activity light does not seem to work, but the status light is solid green.

PowerEdge HDD/SCSI/RAID

I/O Errors with Perc H310

Was this post helpful?