SATA Controller Cards


Recommended Posts

1 hour ago, coblck said:

Would THIS work, i know it only 4 port but thats all i would need

 

Keeping track of the LSI model number and the firmware that is on these cards is extremely difficult.  Some LSI firmware provides a raid JBOD pass-through mode that changes the drive apparent serial number which makes it very difficult for most folks to track which physical disk is actually associated with the serial number that unRAID show in the GUI.  The card must have the "IT mode" software installed on it.  Sometimes the cards can flashed to have the IT mode software.  But you have which cards and which software to use.  (I understand that is not very difficult to do if you have the right card.)  if I would suggest that you read this post and stick to these model numbers.  On E-bay, reputable vendors will often list that the cards support IT mode (because only a very few LSI models have it installed by default).  (Most of their sales are to folks who are building commercial RAID servers and that is their big customer base.) 

 

    https://lime-technology.com/forums/topic/69018-sata-controller-replacement-question-and-advice/?tab=comments#comment-630097 

 

Stick to those model numbers and save yourself a lot of grief.  There are a lot of reputable vendors on E-Bay who sell these LSI and OEM cards that use LSI chips that require cross flashing (a more difficult process) but you will pay a bit more for them since they are the ones which most folks can use and want!  

 

Edited by Frank1940
Link to comment

Ok thanks again alot of thinking and reading to do, i installed Syba SY-PEX40039 - asm1061 chipset 2 port sata card today and plugged in my ssd cache drive and rebooted and all seems good so far but like the idea of these lsi cards 

Link to comment
On 4/14/2018 at 8:40 PM, tucansam said:

 

 

I love my Silverstone cases.  They make great stuff.

 

Having said that, the controller you referenced above looks like the Chinese knockoff LSI I got on ebay last year.  It won't do more than 20MB/s no matter what I do.  And there was speculation that, if its firmware isn't 100%, it might lead to data corruption.

 

I'm not knocking Silverstone, maybe they contracted with the same plant that builds them for LSI.

 

Or maybe not.

 

 

I have moved it to a pci-e 3 x8 slot and it is working great now. I contacted Silverstone to ask them about the issue i had with the ddr 2 motherboard and they replied from a .de domain, indicating them are a German company. They said that the firmware file for IT mode says it is for 9207 but it will work on the 9217 as only raid mode available labelled 9217. Strange why not just label it too match but I guess it doesn't matter. 

 

I am currently doing pre clear on four disks, temps are a bit high but usually i imagine they idle at a much more reasonable temperature. It is a stress test of sorts anyway. 

 

The pre-read stage is showing 220mbyte per second for all four drives. 

 

 

 

Edited by Simontv
Link to comment

Anyone using the LSI SAS 9400-16i card in your unraid server?  This card can drive 16 sata drives and I need more ports.

 

update:  After reading more, this card is way overkill and plus it is NOT on the hardware compatibility list.  I also noticed that the LSI 9305-16i is not on the list.

 

I guess  I will have to go with the LSI 9305-24i  because it is on the list and recommended.

For LSI cables, I see LSI LSI00410 0.6m Internal Cable SFF8643 to x4 SATA.  

 

I'm looking for a recommendation on cables.  I'm looking for the 90 degree facing down sata connections, so the cables drop downwards.  The ones that go straight in make it tight for my side door, they rub against the door.  I have tried 90 degree up cables, but those just make a big mess and the cables are in the way.

 

The only LSI cables I can find are straight, so I need to look at other brands.  Any recommendations?

Edited by Switchblade
Link to comment
  • 2 weeks later...

 

I went with LSI cables for two of the ports and Supermicro cables for the other two.  I hope they are more relieable than the crappy Trip Lite ones that get errors if you just look at them.

 

LSI 9305-24i arrived and will install soon.   After checking the website, I see there is a new firmware release.  While the hw capatibility list says works Out Of the Box, has anyone updated their firmware to this latest version, 16.00.01.00 - released May 20, 2018?   https://www.broadcom.com/products/storage/host-bus-adapters/sas-9305-24i#downloads

 

Trying to decide if I leave the firmware as is, or update to the latest before connecting my drives.

Link to comment

I am not sure of Broadcom history about delivering bug-and-trouble free updates, but I would be feel better about using the card as received to make sure it works (no warranty issues that way) and see if I had a problem with the board in my unRAID server that a firmware upgrade might fix.  

Link to comment

Thanks I agree, that is the safe strategy and normally what I do. That said, when they say there are 12 defects, means you should at least assess if they impact unraid setup.  This is a minor dot lease and after reviewing what this fw update fixes, they are very minor.  I didn't see anything in the change log that would impact performance in unraid.

 

For those that want to see what the 12 fixes changed, see below.  I'm sorry this is a long post, but this way these items will be easy to search on this forum.

 

ReleaseOrder ID: SCGCQ01748201
Headline: Point Release: SASFW_Intruder_Phase_16.0 - 16.00.01.00 IT Release
Release Version: 16.00.01.00
UCM Project: SAS3.5FW_MASTER_DEV
Sub UCM Project: SASFW_Intruder_Phase_16.0
UCM Stream: SASFW_Intruder_Phase_16.0_Rel
Release Type: Point
State: Open
Release Baseline: SASFW_Intruder_Phase_16.0-2018-05-01-16.00.01.00_REL_1525219757@
\SAS_CTRL_FW
Release Date:
Date Generated: May 02, 2018
Defects Fixed (12):

 

ID: SCGCQ01720974 (Port Of Defect SCGCQ01681441)
Headline: (SATA only) Self-test offline immediate status received from drive is not populated to SCSI log sense command.
Description Of Change: In certain firmware condition self-test in progress was reported without parsing read log ext data.
Change is done to parse self-test status from read log ext self-test log output and populate SCSI log sense data
Issue Description: Self-test status queried via SCSI log sense command always reported as self-test in progress even after self-test completed at drive end.
Steps To Reproduce: Trigger Offline Self-test on SATA drive using SCSI send diagnostic command.
Poll Offline Self-test status using SCSI log sense command.

 

ID: SCGCQ01732726 (Port Of Defect SCGCQ01571933)
Headline: (SATA Only) ATA Request Sense Ext command sometimes not send when the drive responds with 'Sense Data Available' bit set.
Description Of Change: Added check to see if 'Sense Data Available' bit is set in the response handler function for ATA Identify command. Removed some checks that were preventing sending ATA request sense
command from some completion handler functions.
Issue Description: Whenever a SATA drive responds with 'Sense Data Available' and 'Error' bits set in its reply Device to Host FIS firmware is required to send ATA request sense ext command to retrieve the sense
data. It is observed that in some cases the command is not sent.
1. When the SCSI command is translated to ATA Identify command (but changed to some other command using Jammer), since SATA drives are not supposed to fail the Identify command a check
to see if the 'Sense Data Available' bit is set was not added.
2. When a command requires drive to send the PIO setup FIS to initiate data transfer after the translated ATA command is sent by the firmware but instead drive responds with device to host FIS
with Sense Data Available bit set, some checks in firmware was preventing it from sending ATA request sense ext command.
Steps To Reproduce: To a SATA drive that supports sense data reporting feature send a SCSI command that results in failure by the drive (or change the Host to Device FIS using Jammer that the drive fails) with
'Sense Data Available' bit set. Observe that for sometimes ATA request sense command is not send by the firmware to get the sense data.

 

ID: SCGCQ01732727 (Port Of Defect SCGCQ01577907)
Headline: (SATA Only) SCSI Write Long command send with out of range block address does not fail.
Description Of Change: Before sending ATA Request sense EXT command check if drive is in automated mode if so change it to non automated.
Issue Description: SCSI to ATA Write long translation changes the drive to automated mode for data transfer but when the drive fails the command with Sense Data Available bit set in Status field, firmware will send
the ATA Request Sense Ext command to the drive to get the sense information, but it needs to change the mode of drive to 'non automated' other wise the successful response of Request sense
command will result in responding to original Write Long command.
Steps To Reproduce: To a SATA drive that supports sense data reporting feature send a Write long command and through a jammer in translated ATA FIS change the LBA value to a out of range value. The command
will complete successfully instead of failing with check condition as expected.

 

ID: SCGCQ01732728 (Port Of Defect SCGCQ01593844)
Headline: (SATA Only) ATA Request Sense Ext command not sent for failed SCSI Unmap command sent to a SATA drive.
Description Of Change: Added code to check the 'Sense Data Available' bit in the function handling response for Data Set Management command send to the SATA drive while translating SCSI Unmap command.
Issue Description: SCSI Unmap command sent to a SATA SSD gets translated to Data Set Management (TRIM) command, if this command is modified such that the drive fails the command with 'Sense Data
Available' bit is set. Firmware is required to send a ATA Request Sense Ext command to get the sense data, which is not being sent.
Steps To Reproduce: 1) Send a SCSI Unmap command to a sense data supported/enabled drive
2) Modify the sector count (In ATA FIS) to a huge value
3) sense data available bit is set in the drive
4) ATA Req sense Data Ext command is not sent to drive by SATL

 

ID: SCGCQ01732730 (Port Of Defect SCGCQ01598756)
Headline: (SATA Only) Incorrect Additional sense code when SCSI Format to a SATA drive fails
Description Of Change: Changed the additional sense code to PERIPHERAL DEVICE WRITE FAULT.
Issue Description: As per Scsi to ATA Translation spec when write operation during Format handling fails then firmware should respond with sense key set to MEDIUM ERROR and the additional sense code is set to
PERIPHERAL DEVICE WRITE FAULT. Currently firmware is responding with sense key set to MEDIUM ERROR and the additional sense code is set to FORMAT Command Failed.
Steps To Reproduce: Send SCSI Format Unit command to a SATA drive with IMMED bit not set in parameter list and when the LBA field (In ATA FIS) modified to a huge value to force a failure during the write
operation , Additional Sense Code is set to "Format command failed" whereas it should be set to "PERIPHERAL DEVICE WRITE FAULT"

 

ID: SCGCQ01732731 (Port Of Defect SCGCQ01599542)
Headline: (SATA Only) ATA Request Sense Ext command being sent after ATA command translated for SCSI Start Stop Unit command fails with Sense Data Available.
Description Of Change: Removed sending ATA request sense ext command in completion handler function for start stop unit command.
Issue Description: For any non - NCQ ATA command sent to SATA drive as part of translation from SCSI to ATA, that fails with Sense Data Available bit set in Status field Firmware sends ATA Request Sense Ext
command to get the sense data. However in case of SCSI Start Stop Unit command the Scsi to ATA Translation specification describes what sense data should be used if any translated command
fails. So there is no need to send the ATA Request Sense command.
Steps To Reproduce: Send SCSI Start Stop Unit command to a SATA drive that supports sense data reporting feature and force it to fail with sense data available bit set.

 

ID: SCGCQ01732734 (Port Of Defect SCGCQ01605100)
Headline: (SATA Only) ATA Request Sense Data Ext command not send to device when it fails translated SCSi Security Protocol Out.
Description Of Change: Set the error status correctly when the error response is received before calling the completion handler function.
Issue Description: On receiving an error response for a non NCQ ATA command with sense data available bit set firmware should send the ATA Request sense data ext command to get the sense data but this is not
happening for Security Protocol Out command because the completion handler function was not being called with correct status in case of error.
Steps To Reproduce: To a SATA drive that supports sense data reporting feature send a SCSI Security Protocol Out command such that the translated command is failed by the drive with sense data available bit set.
Observe that firmware is not sending the ATA Request Sense Data Ext command to get the sense data after the failure.

 

ID: SCGCQ01732736 (Port Of Defect SCGCQ01628346)
Headline: (SATA Only) Incorrect Sense Key and Additional Sense Code set for failed Start Stop Unit command.
Description Of Change: Removed a check that was preventing f/w from processing the error response in completion handler of one of the translated ATA command for SCSI Start Stop Unit command.
Issue Description: During translation of SCSI Start Stop Unit command if any error is reported by the drive the command should be failed with check condition as mentioned in SAT spec. This was not happening in
some cases.
Steps To Reproduce: Send SCSI Start Stop Unit Command to SATA drive with power condition set to standby and force it to reply with error. The command is not failed back to host with check condition.

 

ID: SCGCQ01732738 (Port Of Defect SCGCQ01631836)
Headline: (SATA Only) SCSI Log Select command not failed with check condition when translated ATA command is failed by SATA drive.
Description Of Change: Added check for error response from drive in PIO setup handler function of SCSI Log Select translation.
Issue Description: The callback function handling the PIO setup FIS received when translating the SCSI Log Select command was not checking the error response from drive resulting into failure of the SCSI
command but without check condition as expected.
Steps To Reproduce: 1) Send a Log select command to a SATA Drive
2) Ensure that the drive is kept busy (may be using sanitize command with polling disabled)
3) Observe that SATA drive fails the ATA command but SCSI command to host is not failed with check condition.

 

ID: SCGCQ01742589 (Port Of Defect SCGCQ01715509)
Headline: PL: Chassis Slot Valid Flag is Incorrectly Cleared in Enclosure Page 0
Description Of Change: Set the ChassisSlotValid flag in PL enclosure flags when processing SMP Report General response.
Issue Description: Because SCGCQ01323680 was not fully ported, the ChassisSlotValid flag in enclosure flags was not being set. This resulted in chassis slot valid not being set with Enclosure Page 0 was read.
Steps To Reproduce: Use customer system. Zone in a slot in the enclosure. The slot ID will be wrong.

 

ID: SCGCQ01745857 (Port Of Defect SCGCQ01510418)
Headline: Requested capacity field translations are incorrect for remove element & truncate command
Description Of Change: check added for following
1. If requested capacity upper two bytes are not zero then fail command with invalid field in CDB
2. If requested capacity is 1 then fail command with invalid CDB
3. For values of requested capacity other than 0 and 1, set MaxLBA to requested capacity -1
Issue Description: Based on SAT5r00a remove and truncate command with requested capacity set to 1 should be failed. However current firmware implementation ignored this field and sent command to drive. This
was completing successfully as compared to expected behavior of command failing with illegal request.
Steps To Reproduce: Send remove and truncate command with requested capacity set to 1. This command gets sent to drive and completes successfully.

 

ID: SCGCQ01745933 (Port Of Defect SCGCQ01715677)
Headline: PL: Avoid Duplicate Mid Workaround Does Not Start a TM If Device is Removed and Re-Added
Description Of Change: Fixed a bug where the ResetNeeded flag in firmware is incorrectly cleared when a device is removed in firmware.
Issue Description: If a drive is completely removed in firmware (DMD expires and the driver removes the device), and then the device is re-discovered, the avoid duplicate mid feature will not start a TM like it's
supposed to.
Steps To Reproduce: Pull a drive. Wait for DMD to expire. Re-insert the drive. View the firmware log and see that the avoid duplicate mid feature did not start a TM. 

 

Link to comment
On ‎5‎/‎25‎/‎2018 at 4:32 PM, Switchblade said:

After reading more, this card is way overkill and plus it is NOT on the hardware compatibility list.  I also noticed that the LSI 9305-16i is not on the list.

 

I guess  I will have to go with the LSI 9305-24i  because it is on the list and recommended.

 

Careful! The list is maintained by us, the unRAID comunity.

The cards you see there have been tested by users and found to be working.

If a card is not on the list it doesn't mean it will not work!

If it's tested and confirmed not working, then you might find it in the "Not recommended" list.

 

Link to comment

Reading through this thread. Bought the IOCrest from Amazon after reading about it here on the forum but now I may be having hardware problems and the marvel chipset came up as a potential issue.

 

i really am having a hard time making heads or tails of this thread. I just want to add 2-4 ports to my motherboard in a PCIx1 slot. What should I buy?

 

thanks

Link to comment
37 minutes ago, nickp85 said:

Reading through this thread. Bought the IOCrest from Amazon after reading about it here on the forum but now I may be having hardware problems and the marvel chipset came up as a potential issue.

 

i really am having a hard time making heads or tails of this thread. I just want to add 2-4 ports to my motherboard in a PCIx1 slot. What should I buy?

 

thanks

 

Here are @johnnie.black recommendations. 

 

         https://lime-technology.com/forums/topic/69018-sata-controller-replacement-question-and-advice/?tab=comments#comment-630097

 

 

I have bought two cards (see specs below) on E-bay.  I have always searched for cards that have had the IT mode software already installed on them.  I also select a vendor who has a good reputation.  (It costs a bit more but the peace of mind is worth it to me.)  Each card was about $80US including shipping.  (Try to avoid the cards straight from China as some are counterfeit.  The 'used' cards may be the best overall value.  Most of these were originally in servers built by some of the biggest names  in the computer world and they used only the highest quality parts.) 

Link to comment
1 hour ago, Frank1940 said:

 

Here are @johnnie.black recommendations. 

 

         https://lime-technology.com/forums/topic/69018-sata-controller-replacement-question-and-advice/?tab=comments#comment-630097

 

 

I have bought two cards (see specs below) on E-bay.  I have always searched for cards that have had the IT mode software already installed on them.  I also select a vendor who has a good reputation.  (It costs a bit more but the peace of mind is worth it to me.)  Each card was about $80US including shipping.  (Try to avoid the cards straight from China as some are counterfeit.  The 'used' cards may be the best overall value.  Most of these were originally in servers built by some of the biggest names  in the computer world and they used only the highest quality parts.) 

 

Thank you!  I'm looking through the LSI cards and it doesn't look like any are for an x1 slot.  My motherboard has x16 ad x1 slots only.  I'm using my unraid as a Windows 10 VM with a GPU pass through so if I use the 2nd or 3rd x16 slot, the first one is locked in at x8 I believe.  So I'm trying to use the x1 slot and only need 2, maybe 4 ports for room to grow.  Any ideas on that?

Link to comment
6 hours ago, nickp85 said:

My motherboard has x16 ad x1 slots only. 

If you don't want to use the x16 slot you can used Asmedia based controllers on the x1 slot, e.g. SYBA SY-PEX40039 , they are 2 port controllers though, AFAIK 4 port x1 controllers only Marvell, based on the 9315 chipset and not really recommended, but they do work for some.

 

 

Edited by johnnie.black
Link to comment
7 hours ago, johnnie.black said:

No, I have one myself.

 

Thank you!  I see it on Amazon for under $20.  Only need the 2 ports and have 4 PCIe x1 slots so could technically add more if I need more ports I guess.

 

I have an ASUS z370-A motherboard with 3 PCIe x16 slots.  I was reading the third slot always operates at x4 but is it only expecting a VGA card or could I throw the 4i LSI card in there to get 4 ports out of one card?  Found a used one on Ebay from a company selling stuff from IT shops and it comes with IT mode enabled.

 

Edit: ASUS is useless and could not answer this question nor could I find an obvious mention in the manual.

 

Edit2: Review for the board shows PCIe x16_3 runs at x4 over the chipset.  So I guess I could just stick an LSI 4i card in there and get up to 4 extra SATA ports out of it.

http://www.guru3d.com/articles_pages/asus_prime_z370_a_review,3.html

Edited by nickp85
Link to comment
18 hours ago, nickp85 said:

 

No issues with an asmedia chipset?

 

I haven't had any problems personally, whereas I have had trouble with Marvell and LSI (and the occasional bug with the 4 port PCI-X Silicon Image chips, but otherwise the Silicon Image have been good).  ASMedia are owned by Asus, and they also design the integrated SATA controllers on newer AMD chipsets (Q6 at https://www.anandtech.com/show/11177/making-amd-tick-a-very-zen-interview-with-dr-lisa-su-ceo ).

 

I have even considered making my own 16 port SATA cards using 8 of the ASM1061 chips with a PCIe switch chip onboard for a large commercial project.  It's a shame that they only do a 2 port standalone version.

 

This open source NAS design chose the ASMedia chips - https://lwn.net/Articles/743609/

Link to comment

Very new here. I'm in the process of spec-ing and building a new server to run with UnRaid. I've been looking at the 3Ware SAS 9750 24i4e as my controller card. I'm having trouble determining if this is an appropriate choice. The controller chip is the LSI-2108.

 

I found a very confusing post about maybe possibly having to flash this chip, but frankly, I couldn't make heads or tails of it. Anyone got any ideas?

Link to comment
9 minutes ago, Moose_Flunky said:

The controller chip is the LSI-2108.

That's a RAID chipset, not recommended for unRAID, you want a plain HBA, any LSI with a SAS2008/2308/3008 chipset in IT mode, e.g., 9201-8i, 9211-8i, 9207-8i, 9300-8i, etc and clones, like the Dell H200/H310 and IBM M1015, these latter ones need to be crossflashed.

Link to comment
2 minutes ago, johnnie.black said:

That's a RAID chipset, not recommended for unRAID, you want a plain HBA, any LSI with a SAS2008/2308/3008 chipset in IT mode, e.g., 9201-8i, 9211-8i, 9207-8i, 9300-8i, etc and clones, like the Dell H200/H310 and IBM M1015, these latter ones need to be crossflashed.

 

Thank you! I was really looking for a controller in the 16 to 24 port range, which is why my attention went to the 3Ware SAS 9750. Any recommendations for 16 port cards?

Link to comment
  • JorgeB unpinned and locked this topic
Guest
This topic is now closed to further replies.