Nice Deal $28 - Adaptec ASR-71605 PCI-E 3.0 x8 16 Port SAS Controller

Markess

Active Member
May 19, 2018
534
239
43
I was totally clueless when I used the calculator at LFM To CFM Calculator. I just used the default carrier height and width of 8 inches each. Not sure how to get your numbers.
I didn't know how to approach this either originally. But, I'd seen the "200 LFM" specification with other cards (usually buried deep in the documentation), going all the way back to LSI SAS2008 based ones. Since the specification has been around for a while, for both LSI and Adaptec, others have posted info on how to approach it. And, being both a rule follower and copy cat :), I've been cooling all my HBA/RAID cards since none of my systems have very high airflow.

The full spec as I understand it is 200LFM with the air at an ambient temperature of up to 55C measured 1 inch from the card. As @kiteboarder mentions above, how in the world do you monitor that?

From what I've read, the only component that needs the high airflow is the heatsink. So, if adding a fan onto/near the card for cooling, you only need to calculate airflow to cover the heatsink's surface area. This calculator has a circular air source option so you can try different fan diameters to determine airflow at different LFM: ENGINEERING.com | Air Flow Conversion Calculator . Besides CFM it also gives volume in m3h, which is helpful if you're looking at Noctua or other brands that use m3h instead of CFM.

Some time back, I read in posts elsewhere that, if the 200LFM requirement is for air up to 55C (which is pretty darn toasty), it would follow that airflow could be lower when the air temperature is lower. So, I've usually put a lot less than 200LFM on my cards and haven't had a problem. Admittedly, in my home setup, the sample size is pretty low.

For my ASR-71605 in a 3U chassis that otherwise ran very cool, I put a Noctua 80mm fan in a holder in the next PCIe slot over. Plugged into a PWM header, it only moves about 80LFM at its usual speed, but I had no issues with heat.
 
  • Like
Reactions: klui

TubaMT

Member
Jul 26, 2014
117
22
18
Regarding phantom / duplicate drives in SMART data:

I've attached a few pictures to show what we're talking about. I have 19 drives in my system, but HDD Sentinel thinks it has 24. The adaptec card just gives all kinds of incorrect SMART data to the program.

View attachment 15002

In the attached pics, note I that I actually only have two Micron S630 drives. Picture A shows them as drives #0 and 6. Picture B will show them duplicated as drives #15-16. You'll see lots of other garbage data in picture B.


View attachment 15003View attachment 15004


I would go back to the LSI HBA if they would recognize some of the newer SAS3 HDDs.
Are you also getting any issues that are mentioned in this thread: https://forums.servethehome.com/ind...05-corrupt-data-on-raw-drives-uefi-gpt.29417/

It seems like it only affects systems that are booted using UEFI instead of BIOS and that have larger drive sizes.
 

kiteboarder

Active Member
May 10, 2016
102
39
28
41
Critical firmware bug posted above! UEFI + 8 TB drives or larger = corrupted data

Just another reason to toss out this adapter.

How many "likes" can I give to you, TubaMT? I would likely have never know this firmware bug without your post. Virtual beer for you!
 
  • Like
Reactions: RedX1 and TubaMT

TubaMT

Member
Jul 26, 2014
117
22
18
Critical firmware bug posted above! UEFI + 8 TB drives or larger = corrupted data

Just another reason to toss out this adapter.

How many "likes" can I give to you, TubaMT? I would likely have never know this firmware bug without your post. Virtual beer for you!
Lol! Glad I brought this to your attention! Reading this thread, I thought I could deal with the phantom drive thing and almost bought a bunch of cards. But then found out about the possible corruption bug and that is definitely not something I want to deal with lol.

Btw, what's your server setup like with Windows? Are you using in built storage spaces or drivepool?


Don't believe everything you read.
[Also, note that 4TB drive was also "affected".]
It looks like you were the one talking with the person who was having these issues on the snapraid forums. Are you saying that it was a made up issue and did not occur?

Still, seems like there are some bugs with these adapters for some people. Anybody with these adapters able to recreate the data corruption issue? Seems like everyone is having different experiences with these cards.
 

kiteboarder

Active Member
May 10, 2016
102
39
28
41
Responding to UhClem and TubaMT,

UhClem, are you saying that you have no issues? I just want to understand why you disagree with the person that posted the issue in the other thread. I am not taking sides at all - just want to understand your statement. Obviously I would be very glad to hear there is no issue...

Tuba, thanks for the follow up. I'm using Server2019 with storage spaces / ReFS / RAID1. I am "trying" to protect my data. So far no error messages from Windows, but I'm the first to admit that I haven't religiously scanned the scrub log files. (See ReFS off topic comment at bottom.)

I also have qty 4 of 10 TB individual NTFS drives on this controller. They're nearly full with backup data from other machines. I do not yet know if there is lurking corruption, as I have not yet figured out how to locate which specific file would be at the specific corruption address. And I REALLY don't want to run a "file compare" on 40TB of data. Honestly just haven't gotten around to dealing with this yet. If anyone knows how to figure out which specific file would be at that address, I'm all ears.

I don't mean to hijack this thread. I'll also post in the other thread as well.

Thanks to all!

OT regarding ReFS:

My biggest disappointment from Microsoft is their idiotic defaults for ReFS. In my opinion, both data integrity's "enabled" and "enforced" are incorrectly set. You have to constantly run this:
get-item * | get-fileintegrity
get-childitem -Recurse * | Set-FileIntegrity -Enable $True -Enforce $False

...because there is a bug (my opinion) where files copied from NTFS will not inherit the enable and enforce settings of the ReFS drive's folders or root.

For those bored, you can read these two posts:
ReFS integrity is not on by default

He correctly sums it up after the first text box. ;-)

Follow up post. More stuff broken in Windows 10 that only works in Server.
Microsoft: ReFS is ridiculous

My other headaches with ReFS involve their versioning. Server2019 has a different version than Windows Pro for Workstations. You can't take ReFS drives out of a Windows box and stick them in a Server box. You also can't switch an external DAS from Server 2019 down to a Server 2016. Be very careful if you're going to pull drives and move them. It appears to me that ReFS in Windows Pro is for beta testing.
 

thy.flesh.consumed

New Member
Mar 30, 2017
18
8
3
38
There is no issue.
I don't know, the things that he was saying were pretty convincing.

Anyway, I'm trying to replicate that issue right now. Can't put the controller in service if it has a serious defect like that. Sadly I don't have spare 8TB or larger to throw around, but I do have 4TB and Tomas at SnapRaid forums did say he saw the issue on 4TB drives as well. What I did I connected 4TB to internal controller, filled it with data until there was less than 1MB free space left and created CRC32 checksum file. I then shut down computer, connected the 4TB drive to Adaptec, rebooted machine twice for good measure and I'm currently verifying CRC32 checksum file. I'll let you guys know results of the test in the next 24-48 hours.


P.S.: this controller is kind of funky regarding SMART drive pass through but I'm not certain if it's the controller or the software. I've tested CrystalDiskInfo, HDSentinel, and StableBit Scanner.
Windows Disk Management - no issues with "phantom" drives.
CrystalDiskInfo - cannot see any drives attached to 71605 period.
HDSentinel - sees drives, sees SMART info, but also sees "phantom drives" for certain drives only. To elaborate I see "phantom" copy of ST4000DM000 drive that is attached to Adaptec, but I do not see "phantom" drive for Samsung 830 SSD attached to it. Weird. If "phantom" drives bother you I believe you can hide them from view (not sure if you can pause monitoring for phantom drives as I don't normally use HDSentinel)
StableBit Scanner - correctly displays all drive info and does NOT have "phantom" drive issue. Either it simply doesn't have an issue or it is smart enough to automatically hide them from the view.

Overall given than neither built in Windows Disk Management utility nor StableBit Scanner have any phantom drive issues I'm actually tempted to say that this is HDSentinel specific issue.
 
Last edited:
  • Like
Reactions: TubaMT and Aluminat

ViciousXUSMC

Active Member
Nov 27, 2016
201
99
28
38
I have had a Dell R510 on standby for my next storage server for a while. My Dell R710 has finally kind of gotten full.
I cant recall for the life of me if I already have a LSI card flashed to IT mode inside or not lol.

If not it looks like this should work good?

I see some mention of bugs/issues with EFI but I can build this with Legacy BIOS and it will either be ESXi with passthru to the HBA for a Freenas VM or a dedicated bare metal Freenas box.
 

TubaMT

Member
Jul 26, 2014
117
22
18
I don't know, the things that he was saying were pretty convincing.

Anyway, I'm trying to replicate that issue right now. Can't put the controller in service if it has a serious defect like that. Sadly I don't have spare 8TB or larger to throw around, but I do have 4TB and Tomas at SnapRaid forums did say he saw the issue on 4TB drives as well. What I did I connected 4TB to internal controller, filled it with data until there was less than 1MB free space left and created CRC32 checksum file. I then shut down computer, connected the 4TB drive to Adaptec, rebooted machine twice for good measure and I'm currently verifying CRC32 checksum file. I'll let you guys know results of the test in the next 24-48 hours.


P.S.: this controller is kind of funky regarding SMART drive pass through but I'm not certain if it's the controller or the software. I've tested CrystalDiskInfo, HDSentinel, and StableBit Scanner.
Windows Disk Management - no issues with "phantom" drives.
CrystalDiskInfo - cannot see any drives attached to 71605 period.
HDSentinel - sees drives, sees SMART info, but also sees "phantom drives" for certain drives only. To elaborate I see "phantom" copy of ST4000DM000 drive that is attached to Adaptec, but I do not see "phantom" drive for Samsung 830 SSD attached to it. Weird. If "phantom" drives bother you I believe you can hide them from view (not sure if you can pause monitoring for phantom drives as I don't normally use HDSentinel)
StableBit Scanner - correctly displays all drive info and does NOT have "phantom" drive issue. Either it simply doesn't have an issue or it is smart enough to automatically hide them from the view.

Overall given than neither built in Windows Disk Management utility nor StableBit Scanner have any phantom drive issues I'm actually tempted to say that this is HDSentinel specific issue.
What Tomas at SnapRaid does sound convincing and he posted his troubleshooting of the issue so I am not sure how it could be made up. We'll get another data point from you though. Are you running UEFI or Legacy BIOS? If it is an issue, it seems to only affect UEFI systems according to the post.
 

thy.flesh.consumed

New Member
Mar 30, 2017
18
8
3
38
Well, I couldn't reproduce problem Tomas described at SnapRaid forums. I set up a test box, you can see most of the info in the screenshot I attached. My 71605 has the Flash Module Kit with Super Capacitor, not sure if that makes a difference, but wanted to mention it. Additionally my 71605 was set to RAID Expose RAW. The system booted from a single 120GB Sata3 SSD attached to motherboard controller. Before the test I attached 4TB Seagate ST4000DM000 to motherboard controller and filled it with data until there was 844KB space left, almost completely full. I calculated CRC32 checksum on all the files. Then I moved the 4TB drive to Adaptec 71605 and rechecked checksum. It all checked out, no signs of corruption. I actually tested both Legacy BIOS boot and with UEFI BIOS boot as seen in the screenshot, same results - no signs of corruption.

Unfortunately I don't have spare 8TB or greater drive to play around with. So either there is no problem with 71605's, or the corruption just happened to be in the 844KB space I left unfilled, or the CRC32 checksum didn't catch it (what are the chances of that?). Maybe there is no problems with 4TB drives or lower? But Tomas said he noticed corruption on 4TB drive as well.

All in all my test does make me feel better about the card. It would still be nice if somebody else ran the same kind of test.
 

Attachments

Firebug24k

New Member
Apr 12, 2017
26
17
3
37
I did a bit of testing with this guy, using ZFS 0.8.4 on linux. I made a big pool of 16x4TB drives and copied over about 40TB of data, no issues. I then scrambled up all the drives so they were connected to different ports on the expander and/or different ports on the card, no issues. I swapped over to a second 71605 and it still worked fine. I'm still testing but I haven't run across any problems using this card on linux with ZFS. SMART works perfect as well. I set the cards to HBA in their bios.

They don't seem to be very fast, though, and I can't get them to recognize wide/dual link SAS to my expanders either.

I'm doing my testing on a Chenbro/Slicestor 3448 chassis. I can't really tell if it's BIOS or UEFI, the mobo doesn't give me any hotkeys to get into either.

 

kagurazakakotori

New Member
Mar 5, 2017
21
15
3
50
My card with 4x12T drive seems not affected using UEFI+GPT. I was running the latest firmware, raid mode and boot disabled.

Code:
   --------------------------------------------------------
   Controller Version Information
   --------------------------------------------------------
   BIOS                                       : 7.5-0 (32118)
   Firmware                                   : 7.5-0 (32118)
   Driver                                     : 7.5-0 (59002)
   Boot Flash                                 : 7.5-0 (32118)
   CPLD (Load version/ Flash version)         : 10/ 10
   SEEPROM (Load version/ Flash version)      : 1/ 1
 
  • Like
Reactions: Sleyk and TubaMT

TubaMT

Member
Jul 26, 2014
117
22
18
Thank you all for testing. Super informative and helpful. Seems like maybe the issue that Tomas described was an isolated case since you 3 have not had any problems. Very strange indeed but I am glad that things seem to running fine! I would test it as well if I had one of these adapters lol!
 
  • Like
Reactions: Sleyk

Sleyk

Well-Known Member
Mar 25, 2016
1,040
426
83
Stamford, CT
Heheh, I was holding off to comment since I didn't want to come off that I was trying to convince people to ignore a possible issue with the card.
I have personally attached several 8TB drives, along with some 4TB drives in ZFS, and have encountered zero problems. I also tested some drives in Windows as well. I transferred over 8 terabytes to my pool. (I was transferring my whole Hyperspin arcade setup :cool:)

I don't doubt there might have been something, as I know people don't make stuff up, but I suppose it must have been an isolated incident, or perhaps a configuration or motherboard issue since I have not run into any data corruption issues at all. That being said, perhaps the issue arose from a nondetected corrupted firmware, or perhaps some older firmware "residue" on the card, or perhaps even a corrupted flash? All these things could lead to data corruption. If that's the case, and if people are somewhat worried to try this card, I would just say to maybe do a fresh firmware update upon getting it before loading her up and make sure firmware is up to date and looking good :.)

That being said, I'm glad most others were able to verify that the issue is not currently reproducible, since I really like this card, especially for my Windows boxes!
 

Markess

Active Member
May 19, 2018
534
239
43
One of the ebay sellers is having a sale on these at only $22.30 shipped plus tax.

The only downside is they can't guarantee which bracket you'll get.
I did get my requested bracket type when I bought from them a while back, but YMMV of course.

For anyone looking, they've also got got the SFF-8643 to 4xSATA breakout cables a temporary sale for $6.10 w/free shipping in the U.S.
 
  • Like
Reactions: Sleyk

Sleyk

Well-Known Member
Mar 25, 2016
1,040
426
83
Stamford, CT
One of the ebay sellers is having a sale on these at only $22.30 shipped plus tax.

The only downside is they can't guarantee which bracket you'll get.
You're ok either way, as they use the plain old High and Low profile brackets as all the other typical LSI cards. Most people should have a few spare high and low profile brackets laying around. I have a box of oldie Dell H310 and H200 grey brackets just sitting on my table.

Plus, most chinese sellers will ship you a nice shiny new bracket for about a dollar or two. Or you can pick up one from a US seller for maybe 2-3 bucks.

Also, excellent find! @thy.flesh.consumed, Can I have permission to put this on the front page my friend?