Adaptec 71605H with CentOS 7 Linux issues...

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
I was experiencing some instabilities that I originally thought was platform related or something else, but I eventually narrowed it down to the Adaptec 71605H HBA card. Short version of the problem is that this card would cause OS boot failures during a warm reboot, but never from a cold reboot.

original thread with the problem here: having some instability during warm boot...

has anyone else used this 71605H with Linux or any other OS and experience this type of problem? i'm just wondering if this problem is specific to CentOS 7 Linux, Linux in general, or even in Windows or other OSes?

I'm already on the latest firmware from what I can tell.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
so, i wish i had done my homework a little bit more before buying the 71605H. in my search regarding this hanging at boot time during warm boot or dropping of drives, i ran into several postings online with similar issues. I'm posting this here for the benefit of anyone else who might come along searching for this information:

1) first, back in 2013, someone encountered a similar hang of the 71605H with pm80xx (version 0.1.37) driver in kernel 3.10 (which is the kernel for CentOS 7, coincidentally):

The pm80xx driver hangs in 3.10 with the Adaptec 71605H HBA (Linux SCSI)

that thread eventually lead to a patch:

[SCSI] pm80xx: fix Adaptec 71605H hang · linux-pmfs/pmfs@9504a92 · GitHub

which is great, but I haven't been able to figure out if Red Hat back ported this patch to their 3.10 kernel. the kernel RPM changelog has no mention of it that i can tell.

2) in 2014, there was a report of the 71605H randomly dropping drives:

Fwd: Adaptec 71605H HBA randomly failing to detect any drives at init (Linux SCSI)

the conclusion was not encouraging, basically the person switched to LSI controller and gave up on the 71605H:

Re: Adaptec 71605H HBA randomly failing to detect any drives at init (Linux SCSI)

3) similar report around the same year on Ubuntu 12.04.4, which has kernel 3.11 i believe, so not sure if the above patch was applied at this point or not:

Adaptec pm80xx Driver Drops Drives Randomly

4) the review of this card on newegg makes an interesting remark:

Adaptec Series 7H Family 71605H PCI-Express 3.0 x8 MD2-Low Profile SATA / SAS Host Bus Adapter (HBA) - Newegg.ca

I quote it here in case it disappears:

Great HBA with caveats5/29/2016 3:01:43 PM
Pros: SAS/SATA support, up to 12 drives
Works out of the box with current Fedora, Ubuntu, Debian, Arch Linux, Red Hat , and more

Cons: A fan(not included) is absolutely necessary to keep the card functioning properly and at optimal temps.

Other Thoughts: Make sure to update the BIOS from the adaptec website, this card running the old firmware was pretty buggy. No issues whatsoever after the update.
I've got mine running a BTRFS RAID-6 array like a champ. Forget using a RAID card, HBA all the way!
which happens to coincide with some of the problem reports about drives randomly disappearing or not reliably booting when there's more than 12 drives connected, which is exactly my case (i had 16 drives connected).

So, the symptoms are:

- the native Linux pm80xx driver (version 0.1.37) hangs during reboot when there are more than 12(?) drives connected (for sure 16)
- when it eventually does finish booting, only 14 of the 16 drives are detected (this is exactly what i experienced)

BTW, for CentOS 7.4 and 3.10.x kernel, the Adaptec driver source code does not compile. Also, their binary RPM will not install either, even though they claim it is for CentOS 7, perhaps 7.0 or 7.1, but does not work on 7.4.

My Fedora 26 workstation has a newer version of the pm80xx driver at 0.1.38 with kernel 4.14.6. So, I might try to install Fedora 26 with the 71605H and see if the newer driver resolves the problem. If so, it would suggest that the patch #1 above didn't get back ported to RHEL7 kernel, in which case I might open a bugzilla ticket about it or download the kernel source package to confirm the driver doesn't have that patch... that's my next step.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
well, the patch mentioned in #1 above is already applied to EL7 kernel 3.10.0-693.11.1.el7.x86_64. so that's not going to help.

the problem described by #2, 3, and 4 may still exist and caused by something else...

i'm going to try installing Fedora 26 with the newer 0.1.38 driver and see if that helps.

i'm also going to test with less than 16 drives connected to the 71605H and see if there's some point where the problem goes away as I'm usually able to reproduce it consistently.
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
update:

1) the problem i'm experiencing with this 71605H doesn't depend on the number of drives connected. I tried everything from 12x to 16x HDD, and same exact symptoms: hangs on boot when pm80xx driver is initializing, or completes boot, but randomly 2 of the HDD are missing; this was the case even with 12x HDD, only 10 would show up if it completed the boot process on a warm reboot. and as usual, cold boots or power resets have no problems.

2) installed Fedora 26 with the newer 0.1.38 pm80xx driver and still same exact symptoms.

last thing I might try is using Adaptec's own driver, except out of the box it doesn't compile; i'm going to see if i can fix that easily, but if not this is probably end of the road for me.

if anyone is using the 71605H successfully with Linux, I would really love to hear it because I'm unsure if this is a problem with the card + open source Linux driver, or if it is just the card I have acting strange...
 

BLinux

cat lover server enthusiast
Jul 7, 2016
2,669
1,081
113
artofserver.com
Thread necro...
@BLinux - Did you find a resolution to this issue
i still have that card sitting in a box somewhere. I didn't have time to dig into it further, so I ended up swapping the card out for a LSI HBA, but at the same time, I discovered my 846A backplane had a problem and a crystal eventually fell off the backplane. I eventually fixed the backplane with a new crystal, but never tried the Adaptec card again to see if maybe the backplane was what was causing the issues with the Adaptec. The adaptec card would work perfectly on cold boot, but not on warm reboot.
 

nthu9280

Well-Known Member
Feb 3, 2016
1,628
498
83
San Antonio, TX
Thanks. The ASR-71605 (not the HBA) are now going for ~$79. As you may know these can be set to RAID/HBA mode in BIOS. Wondering if it makes sense to consolidate to fewer cards. But after re-reading the thread, it does not appear so. Limit on 12 drives, driver/fw issues etc.
Have not dug into the specs but I'd imagine ASR-72405, ASR-71685, ASR78165, ASR 71605 and ASA-71605H probably use the same driver and probably have the same issues. I don't have the patience and expertise to do this.
 

kapone

Well-Known Member
May 23, 2015
1,095
642
113
It may not apply, but...I have had no issues with the 7 series with Windows. Everything works exactly as it should, including performance.