MPR Files Posted June 13, 2015 Share Posted June 13, 2015 My 9 disk array has been running fine on Unraid v5.0.5. I installed Unraid v6.0rc6 onto a fresh USB stick, bought a new Pro license (because I'm super conservative and didn't want to erase my v5.0.5 drive which is still working fine). Booted up ok but Unraid can't find 3 of my disks. Coincidentally, there are three disks plugged into my STLab A-520 SATA card. It appears to be the same three disks that are not found. Pop Unraid v5.0.5 back in and all works fine. I'm assuming it's some sort of compatibility issue between the STLab A-520 card and Unraid 6.0rc6. Has anyone tested this combination before ? I've tried to find the firmware version of the STLab card, I can get into Marvell BIOS which I suspect is the STLab card. It indicates BIOS v1.0.0.1018 and Firmware v2.3.0.1053 I've googled for the A-520 firmware and can't find anything helpful. Any suggestions would be appreciated. tower-diagnostics-20150613-1756.zip Link to comment
RobJ Posted June 13, 2015 Share Posted June 13, 2015 Howdy, I almost forgot you, saw your post late last night, looked at the syslog, but was too tired for good thinking or writing. Jun 13 17:45:52 Tower kernel: AMD-Vi: Event logged [iO_PAGE_FAULT device=01:00.0 domain=0x0019 address=0x000000000009ea00 flags=0x0000] It sees the 8 port card, but issues the error above regarding it. This occurs just after identifying the card, and just as it is determining what is connected to it. It appears to make good SATA connections on the 3rd, 4th, and last SATA ports, and sees something on the first SATA port but cannot link to it. It is unable to communicate across the SATA links it makes to the other 3 drives. Do the connections I mention make sense with what is actually connected? So far, I have only one idea - because it's related to AMD-Vi, in the BIOS try turning off some or all of the virtualization settings, just for a test. Edit: forgot to ask for a copy of a v5.0.5 syslog, for comparison. Link to comment
MPR Files Posted June 13, 2015 Author Share Posted June 13, 2015 The A-520 card has 4 internal SATA ports, I have three drives plugged into these. It probably has external ports as well, which are not in use. Not sure what you mean by turn off BIOS virtualisation. Is this in the SATA card BIOS or the motherboard BIOS ? Syslog v5.0.5 attached. Appreciate your help. Syslong_v5.0.5.zip Link to comment
RobJ Posted June 13, 2015 Share Posted June 13, 2015 The A-520 card has 4 internal SATA ports, I have three drives plugged into these. It probably has external ports as well, which are not in use. Never heard of an A-520 card before, so not familiar with it. According to the syslog, it claims 8 ports. It does appear to work fine in the v5 syslog, but not in the v6 log. In both, there's a drive connected to the first, third, and fourth internal ports, and something on the last port, apparently external or on the card. The extra item shows the following - Jun 13 18:06:19 Tower kernel: ata16: SATA link up 1.5 Gbps (SStatus 113 SControl 300) Jun 13 18:06:19 Tower kernel: ata16.00: ATAPI: MARVELL VIRTUALL, , 1.09, max UDMA/66 Jun 13 18:06:19 Tower kernel: ata16.00: configured for UDMA/66 Jun 13 18:06:19 Tower kernel: scsi 16:0:0:0: Processor Marvell Console 1.01 PQ: 0 ANSI: 5 Jun 13 18:06:19 Tower kernel: scsi 16:0:0:0: Attached scsi generic sg12 type 3 So it's a slow speed SATA device, but not a drive, a console? Not sure what you mean by turn off BIOS virtualisation. Is this in the SATA card BIOS or the motherboard BIOS ? In the motherboard BIOS. Link to comment
MPR Files Posted June 13, 2015 Author Share Posted June 13, 2015 Had a proper look at the A-520 card. It has 4 internal ports, of which 3 are connected to my RAID disks and 1 is empty. It has 2 external ports, nothing plugged in. The manual describes jumpers on the card to select internal or external SATA ports. It looks like it's actually a 4 port SATA card with 6 ports of which only 4 can be used at any time. Have also grovelled on the floor with a torch to determine that it's a Marvel 88SE9230 chipset card. I've made no changes to the ports the drives are plugged into or the onboard jumpers. Have searched the Gigabyte motherboard manual, no mention of virtualisation. Link to comment
RobJ Posted June 14, 2015 Share Posted June 14, 2015 Not that I didn't believe you, but what the syslog says is so different, I had to look it up on the web. Picture and manual look just the way you said, with 4 internal ports and 2 optional external ports, switchable. But the syslog clearly says the card is claiming to have 8 ports, and presents them for setup, with a 'Marvell Console' connecting by SATA link on the eighth port! In fact, your system is claiming to have 16 SATA ports, 8 onboard and 8 on the card (you can see the Console on the 16th port in my post above). I have NEVER seen a crazier card! I have some bad news though, about this card. A number of Marvell chipsets, including your SE9230, have a bad bug which renders them unusable in a virtualization environment. And the symptoms and syslog errors are exactly what is showing in your syslog. The only workaround is to turn off the directed I/O portion of virtualization, VT-d on Intel and AMD-Vi on AMD. PLEASE, go through your BIOS settings and look again for it, it may be called something else, like AMD-Vi? Directed I/O? Virtualized I/O? While it's turned on, the attached drives cannot be seen. You may also want to look for a BIOS upgrade for the motherboard. Current BIOS is from 2013. It won't help fix the A-520 though. Link to comment
MPR Files Posted June 15, 2015 Author Share Posted June 15, 2015 SOLVED ! Turned off SVM Mode in Gigabyte BIOS. Retested, same problem (can't see drives plugged into STLabs card). Updated Gigabyte BIOS from F4 to F8 Turned off SVM Mode in Gigabyte BIOS. Retested, booting Unraid v6.0rc6 all ok. It's running a Parity Check now at about twice the speed as it did previously (changed Unraid & BIOS, so not sure which one helped). Chewing through the Parity Check at over 100MB/sec, estimated 10 hours rather than over 24 hours previously. Updated diagnostics file is attached. Thank you very much RobJ. If you lived in New Zealand, I'd buy you a beer. tower-diagnostics-20150615-1425.zip Link to comment
RobJ Posted June 15, 2015 Share Posted June 15, 2015 That's great! And thanks for the final diagnostics files, happened to be my first chance to use them to compare a 'before and after' repair. I've posted a Defect report, because I found what appears to be a patch that fixes the problem. I'm not at all sure it can be integrated into a future unRAID, but at least it's there to warn other users. I suspect there may be other v5 users who upgrade to v6 and suddenly lose a bunch of drives. You have the distinction of being the first, I believe! But perhaps this will help us more quickly solve the problem for others. Link to comment
Recommended Posts
Archived
This topic is now archived and is closed to further replies.