July 27, 201114 yr Hi All, I am finally getting around to migrating my array to 4.7 Final. Seem to have an issue with a drive and am seeking advice. I have had intermittent problems with the array when writing large volumes. It will become unresponsive and requires a hard reboot. I have done various things (short of core hardware replacement) to try and diagnose, including disk checks. Log capture reveals absolutely nothing prior to any crashes, so they have been no use. Given I rarely write to the array, I have not exhaustively looked into this. Nevertheless, tonight following a failure during writes, I finally had enough and decided to upgrade to 4.7 and see if this resolves the issue (I know I should have done this ages ago). I was running 4.5.6 and had a valid config, all drives green. I have copied the required files to my USB drive (bzimage, bzroot and memtest for a 4.5.x to 4.7 upgrade). No changes to the hardware config at all, but now I get a red disk when the array boots. Below is the disk screen on boot with the problem drive. It is reporting an incorrect size. Unraid seems to think this disk is new, probably based on the size descrepancy. So the question I have - is this something to do with the 4.7 upgrade, or a good indication that this drive is likely an issue and should be replaced (note the array starts under 4.5.6). Heck this sucker might even be the cause of all my issues. If I should replace the disk, should I roll back to 4.5.6 and replace it from there, then upgrade to 4.7? Or just replace it and let a parity sync take care of it under 4.7? Syslog also attached for reference. Any advice would be appreciated. Thanks in advance! Hardware Mobo - Gigabyte GA-EP35-DS3R 4Gb Ram - Recently Memtested SuperMicro 8 Port Sata PCI card 16 SATA Drives Total No Cache Drive syslog-2011-07-27_-_4.7_upgrade.txt
July 27, 201114 yr Welcome to the world of HPA. Do a search on the forums about it and you should find a lot of info on it. You will need to check the BIOS of your motherboard to see if HPA/Backup BIOS to HD/BIOS Restore/etc can be disabled. I would unplug all drives, update the BIOS, set defaults on the motherboard, and then check to see if HPA is OFF by default. If it is not then you could have problems later when the CMOS battery dies. You will have to remove HPA from that drive and or change your motherboard.
July 27, 201114 yr Author Nasty. Seems at least I don't have HPA on my Parity drive, so that is something. So solution should be as follows (searching forum for specifics): 1 - disable this BIOS feature and ensure mobo is NOT enabling it by default (through updating bios and double checking default is off) , or replace mobo 2 - remove the HPA from the drive by running the relevant hdparm -N command 3 - reboot, assuming all green, run parity. If not, um...seek further advice.... Sound right? Assuming the drive then mounts and the array starts, is there any possibility I'll have issues with parity, given it's a non-parity drive? Should I only run a "Trust my Parity"?
July 27, 201114 yr Yes, that is correct. I suggest following the steps laid out in the HPA thread. If you have any follow-up questions, just post them here.
July 29, 201114 yr Author Hey Guys, I think I have this under control, but I just would like a sanity check before I proceed. I downgraded to a BIOS I know does not support HPA (according to Gigabyte F4 added the functionality, so I flashed to F3). I have checked and rechecked the BIOS (including advanced settings), and can find no reference to anything HPA related. I booted, confirmed the correct drive and ran the necessary hdparm command. It worked (which is probably a good sign the BIOS isn't trying to set it on boot). Now I reboot and the drive is showing as the correct size, but blue. Unraid understandably thinks I upgraded the drive with a larger one. I ran a reiserfsck on the drive, which returned no corruption issues. So, do I now just run a 'Trust My Parity' as detailed below, and cross my fingers? http://www.lime-technology.com/wiki/index.php?title=Make_unRAID_Trust_the_Parity_Drive,_Avoid_Rebuilding_Parity_Unnecessarily Cheers!
July 29, 201114 yr Hey Guys, I think I have this under control, but I just would like a sanity check before I proceed. I downgraded to a BIOS I know does not support HPA (according to Gigabyte F4 added the functionality, so I flashed to F3). I have checked and rechecked the BIOS (including advanced settings), and can find no reference to anything HPA related. I booted, confirmed the correct drive and ran the necessary hdparm command. It worked (which is probably a good sign the BIOS isn't trying to set it on boot). Now I reboot and the drive is showing as the correct size, but blue. Unraid understandably thinks I upgraded the drive with a larger one. I ran a reiserfsck on the drive, which returned no corruption issues. So, do I now just run a 'Trust My Parity' as detailed below, and cross my fingers? http://www.lime-technology.com/wiki/index.php?title=Make_unRAID_Trust_the_Parity_Drive,_Avoid_Rebuilding_Parity_Unnecessarily Cheers! Since unRAID thinks you have installed a larger disk, I'd just let it re-construct the disk to itself. That way, it will have the proper size file-system too. If you do not trust your other drives, then the trust parity might let you skip that step, but you have other issues in regaining the trust of the array anyway. Joe L.
July 29, 201114 yr The Realtek 8111B may be the source of the write problems. You can install an Intel GigE PCI card or try the latest 5.0 beta.
July 30, 201114 yr I had this problem, the motherboard in my server doesn't use HPA but the hard drive in question was an old storage drive I used on a windows pc, which had an HPA on it. I had always ignored it since it wasn't an issue but the upgrade to 4.7 gave me the same problem. I just used the hdparm command to remove it and check it is correct, let unraid rebuild the drive (which was showing up as blue), and am now running a parity check which should fix the space left over from the HPA. I would not do the trust parity step you were talking about. Let unraid do it's thing. Rebuild, then parity check.
July 31, 201114 yr Author Thanks all, pretty much followed that advice and let Unraid rebuild the drive itself. All good now, array is up and all data present. I have a feeling the HPA was added to the drive when I upgraded the MB BIOS a while back. As mentioned above, the newer BIOS added the Xpress Recovery feature, so it probably added it to the drive on the next boot. I also just dropped in an Intel PCIe NIC. Hopefully my issues with the array becoming unresponsive on large writes goes away now. Thanks!
Archived
This topic is now archived and is closed to further replies.