January 18, 201214 yr Hi, I recently added a PCI SATA controller and another HDD to my array and have been experiencing constant server crashes since. Everything seemed to be ok, with the pre-clear operation successful, the drive added to the array, and the rebuild/parity operation complete. However a few hours later the server became unresponsive (was still powered up but unable to ping). A hard reboot gets it going again, but seemingly only long enough for the array to start, before it becomes unresponsive again, and I have done this 3 times now. At first I thought it may be the power supply, but leaving the power connected to the new drive and removing the data cable allows the server to boot and remain responsive (with the the drive in question Missing). I have changed PCI slot for the controller (to rule out IRQ type issue), but it doesn't seem to help - not that I really expected an issue, as it seemed to work fine during the pre-clear/array rebuild operation. I have attached the log, and would appreciate any help: unraid_log.txt
January 19, 201214 yr Well, the one thing that jumps out at me is your change in hardware (which you are aware of as well). The SATA card is likely bad or incompatible with UnRAID (what is it? chipset does it use, etc?). Get another SATA card (get one that is on the list of compatible hardware on the wiki/faq/forums) and see what happens. It could be the hard drive, have you run pre-clear script on it?
January 19, 201214 yr At first I thought it may be the power supply, but leaving the power connected to the new drive and removing the data cable allows the server to boot and remain responsive (with the the drive in question Missing). My first guess is you have 9 drives, most non-green, on a marginal power supply. 9 * 3 Amps = 27 Amps + 4 or 5 Amps more for the motherboard and fans = 31+ amps needed on a single 12 volt rail. Most multi-rail supplies only have a single rail used for ALL the disks AND the motherboard. It typically is rated at 16 to 18 amps max. The other rails are dedicated for the CPU and PCIe video cards. What specific make/model power supply are you using? Unplugging the data cable would result in the drive not being spun up and not having the disk heads load, resulting in less power used. (possibly being just under the threshold needed to boot your server)
January 21, 201214 yr Author Hi, Yes, 9 drives (7 data, parity + cache) but 'most' are green. "Newest" drive is actually my old parity, and is a 1Tb WD caviar green. The new parity is a "Seagate SATA3 2TB Barracuda 64mb Cache" drive. Power supply is a Huntkey Green Power LW-6550HG which, despite the naming convention, appears to have a max total power draw of 550W. I can't check the controller chip at the moment, but I did search the forum before purchasing and it appeared to work for others (with the only issue being IRQ). I had thought that since it completed the rebuild/parity operation it must be ok...?
January 21, 201214 yr Power supply is a Huntkey Green Power LW-6550HG which, despite the naming convention, appears to have a max total power draw of 550W. that PSU is dual +12V rail with about 16a on rail one. while it is a 550, it bottoms out at about 330watts
January 21, 201214 yr The power supply is not appropriate for unRAID. See here: http://lime-technology.com/forum/index.php?topic=12219.0
January 21, 201214 yr Author Looks like I can get a NEW Corsair cx-500 (one of the recommended models in the link provided above) for about $98 delivered... ...or a USED Corsair HX 650W Modular PSU for a lot less... Is one a better value proposition over the other?
January 21, 201214 yr Author I do also have in my posession a Vantec Ion2 van500n power supply, which looks like it's a dual rail, 20 amp power supply: "DC Output +3.3V 25A, +5V 25A (Max Combined Wattage 160W) +12V1 20A, 12V2 20A (Max Combined Wattage 384W) (Max Combined Wattage 484W) -12V 0.5A, -5Vsb 2.5A (Max Combined Wattage 20W)" Could this work if the drives are appropriately shared between rails?
January 21, 201214 yr I would get the new Corsair. Then you can be certain that power issues are gone.
January 27, 201214 yr Author Thanks all for the suggestions, I've now purchased a new Corsair CX-500 and will post progress once it arrives.
February 8, 201214 yr Author PS finally arrived and installed, the data rebuild was successful and I was playing back media during the process so thought everything fine... then I triggered a parity check and the system crashed again... If it's the controller, how is the data rebuild able to succeed?
February 8, 201214 yr Welcome to the club.... mbryanr and I both have similar issues. mbryanr constant crashes were caused by poorly mounted CPU fan. I'm still trying to fix mine. So far, it wasnt' the PSU or RAM. I too was wondering about the controller, but then my server just crashed again w/o any drives attached! Maybe some something related to the ethernet is telling the server to shut down?
February 8, 201214 yr Author Not sure this is a club I want to join... If it was a poorly mounted fan and the crashes were intermittent, does that mean it only occurred under heavy load? I'd have anticipated a parity check to be pretty hard on discs and RAM, but not CPU. When replacing the PS I checked all the case fans, so I'll check the CPU cooler too when I get home.
February 8, 201214 yr Only under heavy load (ie parity check) did it shutdown http://lime-technology.com/forum/index.php?topic=18224.msg163398#msg163398 I followed that up with shutdowns only when all disks were spun down http://lime-technology.com/forum/index.php?topic=18329.msg164152#msg164152
February 8, 201214 yr Author ok, well now I really don't know WTF is going on... CPU is an AMD 1640le low power chip. I removed and reconnected the HSF but no change. Barely warmer than ambient though which seems odd... So I added more RAM but same problem. Watched it POST, watched unraid load, get to the logon prompt and I get only a second or so before keyboard stops responding (I could only type "ro" before it locked up). So also not sure it's the parity check (or associated load) at fault. I also don't think it's the SATA card because I just unplugged a (SATA cable from a) drive on a different controller (i.e. not the new drive or controller), and it's been up and stable for a while now (but obviouisly witha drive missing)... Possibly an IRQ issue, as my system only seems to cope with this number of drives if one of them is PATA? ...but then, why did the rebuild successfully complete?
February 10, 201214 yr WOW... i finally figured out why my server kept crashing. It was a faulty POWER button. The crashes were absolutely random reboots. That's why non of the efforts to isolate the CPU, PSU, RAM,HD's, etc. didn't work. Fortunately the Biostar TH67+ mobo has an onboard button for power on and restart. I disconnected the power on, restart, and LED wires from the mobo and have not experienced a random reboot since. Maybe this will help.
February 11, 201214 yr Author Hey, thanks for the suggestion - that hadn't really occurred to me! Unfortunaltely I can't really confirm now as I think in my case the mobo is cactus... the HSF had goen from slightly warmer than ambient to cold which made me think the CPU was fried, and after shutting it down to reseat the HSF the lights on the keyboard and fans would flash but the system wouldn't even POST (or beep despite a mobo speaker). So I swapped in another CPU - no good. Changed the RAM - no good. Changed the PS - no good. Removed all add-on cards - no good. Disconnected the HDD's - no good... Replaced the mobo and CPU with those from an an old intel workstation and have my array up and running stabily with parity completed over night (minus the new controller and HDD). Have just re-added the new controller and HDD, but still running preclear, so will likely be another couple of days before I know if it's stable with the expanded array.
Archived
This topic is now archived and is closed to further replies.