6.2.rc4 - UnRAID not seeing drives with 2xSASLP-MV8

Hello All,

Long time lurker, first time [finally] have the equipment to run this.  Before making a full purchase, i wanted to do a dry run and I am having lots of problems.

• I got the USB stick to boot
  
• It worked the first few times I booted up
  
• I configured some stuff in the WebGUI and now it's fudging up badly
  

I then just formatted the USB and started over, but it seems that didn't help at all.  I took all my HDD out and just boot UNRAID usb stick - it's still doing some type of panic because I can't seem to get access via Telnet OR webgui.

I noticed when I tried accessing the WebGUI once the login prompt comes up, the screen displays a bunch of crap:

[<ffffffffffff814685ff>] ? ata_eh_reset+0x475/0xa67

and a bunch of other services.  I don't quite know how to get the log for you since I can't seem to access it at all.

Please advise this newbie.

-update- Screenshot via ipmi interface

-update- Specs

-update- Include syslog.txt

SuperMicro x8dtl-3f 36bay chassis (LSI Firmware Update to 1.33 P21)

48GB DDR3 ECC memory

1x5TB drive populated (for testing, will have a variety of 20 drives if I can get it to work)

2xAOC-SASLP-MV8 (no firmware modification)

-Final Update---> Resolved!

First off, let me begin with... I'm an ignorant MF! I thought it would be easy plug and play, but with servers and raid controls and expanders involved - I clearly did not understand the concept before attempting at this.

Thanks to a gentleman at Supermicro (DL) for his 15 minute call and educating me on Expanders and raid requirements (in my native language - not english).

I initially had installed my cards in and hooked up all the mini-sas connectors to every port on the backplane.  I did not realize that my server had an expander backplane and it only required 1xMiniSas connector on the front (24-bay) and 1x to the rear (12-bay) to be connected to 2 of my cards.  I had initially had all 3 of my front expander backplane sas connection hooked to two of the MV8 - which worked fine, but as soon as I had the third/fourth connection from the rear backplane expander - it fudged it up inside of unRAID.

With the knowledge, I unhooked everything and connected one sas connection from one h310 sas-port to the front backplane. I then did the same to the rear backplane on a separate H310. The technician said I could connect both to one card; however, the H310 only supports up to 32 devices and I would not be able to see my full 36.  After doing so - I booted the server up and no hangs or crash.  Everything seems to be running perfectly!  I did not test with the MV8s, but I imagine it was due to my lack of knowledge is why it did not work correctly.

Thank you all for your help with this.

Note: I did reflash the H310s and placed tape over the B5 and B6 pins.

syslog.txt

The AOC-SASLP-MV8 is a raid controller. Som raid controllers even in JBOD mode if they have one don't expose the correct raw access to the drive that unraid requires.

Could be your issue here, I'm sure someone with one of these cards will chime in.

Cheers,

David

Thanks for your reply. It is my understanding that we have lots of users who uses these cards and has had no issue. It was one of the primary motivator of why I went with this model.

I will try unplugging the cards tomorrow and see if that bypasses it. In the interim, if anyone has clues or had ran into this-I could use your feedback.

Thanks!

Sent from my iPhone using Tapatalk.

Try also disabling the onboard LSI controller.

Thank you both for your suggestions.  It helped with diagnosing.  Here's what I found out and I could use further help.

I had placed 1xMV8 in a pci 1.0 x4 slot, while the other one was in a pci 2.0 x4 slot.  Testing with each one by itself - Unraid came up just fine.  Testing both cards in those slots, bombed UnRAID.  I fixed this problem by moving the PCI 1.0 to a PCI 2.0 x8 slot instead.

Question 1: Is there a problem with running the two cards on different types of PCI version?

After getting the raid cards & onboard LSI reflashed raid to work with UNRAID, but no HDD - i started populating the HDDs.  I filled up the LSI with all HDDs, while keeping the MV8's empty - this worked fine.  All drives came up and UnRaid also came up.

I started populating HDDs to the two MV8s and this is where it started having problems.  Through some trial and error with swapping the hdds (keeping the same 2 hdds for control) - I started going 2x2 slots to see if the backplane was the problem.  The two drives worked on all slots - so that ruled out backplane.

I started looking at the drives and I think this is where the next issue is, but am unable to come up with an answer.  I have a Toshiba 5TB 7.2k 128MB cache and a 4TB WD RED.  Whenever both (or each) of these are populated into the MV8s, UNRAID would crash immediately during the checking of the drives (to see if they exists phase?) The error is in my screenshot, exactly as the same error i received when I had posted my OP.

Question 2: Are there any known problems with the latest stable release that might affect these drives?

Question 3: Am I doing something wrong or misconfiguring something that may contribute to this problem?

I have a couple of the MV8 in my system that re fully populated with a wide variety of drive models and encounter no problems.

One issue I found with the MV8 is that it is easy to get them not perfectly seated in the motherboard.  It is easy when screwing in the backplane to put some torque on the motherboard connector.  The symptoms were similar to what you describe in that the system would crash in an unpredictable way under load.  Sorting out the seating into the motherboard has completely resolved this issue.

Question 1: Is there a problem with running the two cards on different types of PCI version?

These controllers are PCIe 1.0, so it shouldn't make any different what gen slot they're in, they'll always run at 1.0.

The SASLP is one of the most used controllers with unRAID, only other thing I can think of is to look for bios updates, for the board and the cards.

You could also try v6.2, it uses a newer kernel with newer drivers, it can make a difference.

Also, unrelated to your problem and you probably already know, the onboard LSI only supports disks up to 2TB.

While I think of it, have you done a CTRL-M during the BIOS boot sequence to get into the BIOS settings for the MV8 card?  That way you can check that the drives connected to it are being seen by the card, and that they are NOT configured to run in RAID mode.

Question 1: Is there a problem with running the two cards on different types of PCI version?

 

These controllers are PCIe 1.0, so it shouldn't make any different what gen slot they're in, they'll always run at 1.0.

 

The SASLP is one of the most used controllers with unRAID, only other thing I can think of is to look for bios updates, for the board and the cards.

 

You could also try v6.2, it uses a newer kernel with newer drivers, it can make a difference.

I am not new to computers and what you said  was what I had understood, so it was odd when it wouldn't work under both pci 1/2.  I will try new kernel.

Sent from my iPhone using Tapatalk.

Also, unrelated to your problem and you probably already know, the onboard LSI only supports disks up to 2TB.

Johnnie,

Yup. Found that out when i placed a 3TB and only saw 2.2. Not a problem, i just wont put anything past 2TB on the LSI.

Sent from my iPhone using Tapatalk.

While I think of it, have you done a CTRL-M during the BIOS boot sequence to get into the BIOS settings for the MV8 card?  That way you can check that the drives connected to it are being seen by the card, and that they are NOT configured to run in RAID mode.

Itimpi,

This was one of the first things I did was ctrl+c for lsi and ctrl+m for mv8 to disable raid, shorten the detection timing to 10 seconds, and disable int 13.

The drives are detected by the controllers, the problem exists after UnRAID booting process starts and during drive detection period within UnRAID.

Sent from my iPhone using Tapatalk.

Question 1: Is there a problem with running the two cards on different types of PCI version?

 

These controllers are PCIe 1.0, so it shouldn't make any different what gen slot they're in, they'll always run at 1.0.

 

The SASLP is one of the most used controllers with unRAID, only other thing I can think of is to look for bios updates, for the board and the cards.

 

You could also try v6.2, it uses a newer kernel with newer drivers, it can make a difference.

Good suggestion Johnnie.  I was able to get the random stalls to keep going until login prompt with the latest RC4 for 6.2.  Once it loaded, i was only able to see the drives from the LSI (9 drives) - while the other 9 that were on the dual MV8 did not appear.  I'm in the process of trying to figure out what the deal is with the MV8s.

Questions I wanted to ask you all:

1) Can a bad hard drive produce a kernel panic, halting the booting/loading of UnRAID process?

2) Can bad memory do this as well (i'm in the process of testing all the memories)

3) Any suggestion on the MV8 to troubleshoot it further? I'm thinking of reflashing it - but a bit hesitant since they work individually, just not together.

Thanks!

P.S. I'll post a diagnostic zip file once I get my memory testing completed.

-update-

Answer 2: all my memory passed after I had reseated them. But it was bad at first and the processes did not stall or stopped.

Answer 3: i went ahead and reflashed both cards individually and matched up configs. For some reason if both cards are installed, but one card had RAID function on, during card initialization period - it does not show that RAID was on on the card(s).

I wanted to give an update as well as provide a log so I could get some help.  It took me a while to figure out to get the log exported.

-I found that when I had both my MV8s installed, even though one did not have RAID mode disabled - it still showed both adapter as JBOD.  Luckily I noticed it when I went to trying single cards.

-Both cards are now properly flashed to 3.0.0.21 with RAID5/RAID FEATURE disabled.  Both cards settings also matched.

-I cleared the CMOS on motherboard and started enabling/disabling different features to see if it worked

-------Still had problems after the above steps, but UnRAID did load up to the prompt once the MV8s were flashed and reconfigured-----

At this point, I am still having trouble with UnRAID:

1. Gives me logon prompt (3/4 times)

2. WebGUI never comes up

3. I do not see any disks connected other than USB when I run command to see what disks are available

I am at a loss and hope that one of you guys catch something in the log to help me along with this issue.

Thanks!

syslog.txt

This is almost certainly a hardware problem with that motherboard/bios with that controller card.

I have seen this exact issue with other motherboards and other controllers in the past.  A few years ago we once placed an order for something like 100 Rosewill 4-port sata controllers, where we needed to use 2 per motherboard for one of our server products.  We went with Rosewill because the one we tested worked fine and it was half the price of an Adaptec controller we were using.  Long story short: when 2 of those controllers were plugged in, one of them would not be recognized.  Didn’t matter which PCI slots, and we even tried different motherboard, but eventually we got confirmation from Rosewill, “yeah that doesn’t work”, and we RMA’ed all 100 of them.

A workaround would be to keep one of the SASLP-MV8’s and use a different vendor/model controller to replace the other SASLP-MV8.

This is almost certainly a hardware problem with that motherboard/bios with that controller card.

 

I have seen this exact issue with other motherboards and other controllers in the past.  A few years ago we once placed an order for something like 100 Rosewill 4-port sata controllers, where we needed to use 2 per motherboard for one of our server products.  We went with Rosewill because the one we tested worked fine and it was half the price of an Adaptec controller we were using.  Long story short: when 2 of those controllers were plugged in, one of them would not be recognized.  Didn’t matter which PCI slots, and we even tried different motherboard, but eventually we got confirmation from Rosewill, “yeah that doesn’t work”, and we RMA’ed all 100 of them.

 

A workaround would be to keep one of the SASLP-MV8’s and use a different vendor/model controller to replace the other SASLP-MV8.

Thank you for offering a response to this issue.  I am in contact with Supermicro and they had me troubleshoot a few things.  One of the troubleshooting step was to remove all drives to see if the MV8 would cause issues booting up.  The result of this test was that UnRAID did come up without any problem (as far as i can tell).  The WebGUI was accessible and responsive.  This is why I do not think it is a hardware issue (the cards or motherboard), it may be a hard drive problem?

I'll continue to troubleshoot with Supermicro as well to see if it leads to anything else.

Thanks!

-update- I am attaching a second syslog.  This syslog was pulled from the WebGUI.  I am dual booting with mv8 and 4 hdd that are attached.  UnRAID does not see the drive and the WebGUI took a while to pull up.  I'm hoping this syslog will be able to pinpoint some problems.

-update- also including diagnostics

tower-syslog-20160829-1824.zip

tower-diagnostics-20160829-1838.zip

Supermicro support has asked me to load windows and see if the same problems are occurring as their best guess is that the OS is not properly loading the drivers for the secondary MV8 card.

I'll report back my result once I have Windows Server loaded.

I doubt it is something as simple as that.  I know that both myself and my brother have 2 of these cards in our unRAID servers, without issues and I am sure that there are lots of others that are in the same situation.