March 31, 201016 yr I think the motherboard I used in my unRAID Pro server build has some power issues. It was an old board that had been in the closet for a while. Drives keep disappearing from the array after reboots. While unRAID powers down the system does not turn off completely. The fans still run on the case and processor. I was wondering if I am able to change out the motherboard but still keep all my data?? I was either going to try the motherboard from my old video server or buy one of the motherboards used in the Lime-Tech builds. Any help would be appreciated. I'm using an Adaptec 1430SA and 2x Promise SATA300 TX4 cards on an ASUS P5K-E motherboard.
March 31, 201016 yr Sure. Just keep a backup of your configs and take note of drive assignments. Do a parity check before the change, too. unRAID's kinda like those Linux LiveCD's. It looks for drivers on boot and loads the whole OS to the RAM.
March 31, 201016 yr Take a screen shot of the first page so you know how to assign the drives. The most critical assignment is your parity drive. if you accidentally assign a data drive to the parity slot, you will over write and destroy your data.
March 31, 201016 yr Author I have kept good notes on exactly which drive is which. I labeled each drive and kept a list with the serial numbers. It's just the board seems to have something wrong with it. I'm either going to use an existing board I have but I lose 2 SATA ports or buy a new board. I have two new boards in mind that I need to check compatibility on. XFX MI93007 AS9 SUPERMICRO MBD-C2SBC-Q-O GIGABYTE GA-945GCM-S2C (existing board) I'm not sure I can get all the drives recognized again to run the parity. Is it absolutely necessary to run the parity check before the motherboard swap? I realize that if a drive fails I'll be out of luck. Thanks for the help guys!
March 31, 201016 yr I wouldnt try to run a parity check with an unstable system. Look through some HPA threads before trying the Gigabyte board. Are you sure it is the MB and not the PSU? Assumming the MB, TX4's and adaptec sata ports are populated that is a lot of drives.
March 31, 201016 yr Author I'm changing over to the SUPERMICRO MBD-C2SBC-Q-O motherboard. It has a dedicated PCI-Express x4 slot for the Adaptec card. I'm using a Seasonic 600 watt power supply. With the Supermicro board I can get rid of the video card's power draw. I'll have the following SATA ports full: 6 on the motherboard 4 on the Adaptec 1430SA 4 on the Promise SATA300 TX4 3 on the second Promise SATA300 TX4 Total of 17 drives. Possibly going to 18, not sure if I want to go with a cache drive or just another data drive. I did one of those online power supply calculators and got a minimum recommendation of 499 watts.
March 31, 201016 yr Author It's a Seasonic S12-600. Looks like dual independent +12V output lines. I don't know that much about power supplies so I'm not sure what the real differences are between the single rail and dual rail PSU's.
March 31, 201016 yr Thats a good brand PSU and is an SLI approved supply so should have a hefty 12v rail. That is a lot of HDDs. If you have another PSU to swap in I'd try that before condemning the MB. It sounds more PSU related than MB related. Having a duff MB shouldnt leave the fans spinning. A duff motherboard shouldn't cause the HDDs to be not recognized on a reboot. Both sound more like a collapsing or over stressed PSU.
March 31, 201016 yr It's a Seasonic S12-600. Looks like dual independent +12V output lines. I don't know that much about power supplies so I'm not sure what the real differences are between the single rail and dual rail PSU's. One of those rails is probably dedicated to the CPU and motherboard, the other to the disks. You might have only half the capacity you think you have from the power supply. I agree, it is a lot of disks for a single rail of a dual rail supply.
March 31, 201016 yr like all current seasonic cores, the psu is a single rail design. the only multi rail it has is on the sticker. there is no internal multiple rail fusing. the reasoning for this is to get approval for atx specification which states 12v can only be 18A, hence multiple rails. applies to all seasonic cored psus, ie the corsair hx range also. my mai pc runs of a psu which has two 12v outputs in 5 fused railes, no issues. its 1kw and i use a peak of 750W under full load, with 24 hdds (and the rest of the pc).
April 2, 201016 yr Author WOW, I just went through absolute hell with this unRAID server. I swapped the motherboard and then for the last 2+ hours I've been trying to get my drives back. Well I finally did it. I stepped it back to the new motherboard's main 6 SATA ports and I saw my drives again. Then tried to add the Adaptec card but the unRAID OS would stall during boot. It eventually booted but I couldn't see the drives. I then moved the Adaptec card to the x16 slot on the new supermicro motherboard and it now works. I've got 12 drives up and running parity check now. My shares are back and it appears I haven't lost any data. Once the parity check is done I will add the remaining 5 drives to the array. This has been one of the longest computer projects I've ever done. Mostly due to copying over my video files to the new unRAID server. Some technical issues slowed it down a bit as well. I see the light at the end of the tunnel now. Hopefully I'll be done with this by Saturday and I'll be able to put it into production.
April 2, 201016 yr the main reason for flakey booting problems with pcie cards is theyre not seated correctly. they maybe slightly ajar causing the most funny of errors, for example, it may boot and and load bios but not detect drives, but when seated all teh way in failes to boot (cough ga ep45 cough)
April 6, 201016 yr Author So I'm just about good to go. All my drives have been added and everything seems to be working ok except I get errors on disk 1 and my server likes to go to sleep for some reason. I have no clue why it's going to sleep. I cannot wake it except for a full shut off and turn back on. I'm going to see if I have a keyboard connected if a key stroke will wake it. This is not a good solution. I need this thing active 24/7. I'm fine with the spin down of the disks just want it available at all times.
April 6, 201016 yr So I'm just about good to go. All my drives have been added and everything seems to be working ok except I get errors on disk 1 and my server likes to go to sleep for some reason. I have no clue why it's going to sleep. I cannot wake it except for a full shut off and turn back on. I'm going to see if I have a keyboard connected if a key stroke will wake it. This is not a good solution. I need this thing active 24/7. I'm fine with the spin down of the disks just want it available at all times. About the server going to sleep - I had that problem for awhile. It turned out to be one of two issues: Having duplicate files (accidently self induced) in the same shares but on different drives and/or letting the directory cache script cache all directories. It was causing all of my ram to be consumed and unRAID would go wonky. To fix that I set the directory cache to skip a bunch of deep directories and set it to 3 deep (instead of like 15 or so). And secondly I cleaned up the duplicates which there weren't actually so many but I think it makes unRAID unhappy to have them. About the errors on disk1. I may have mentioned that I have a disk where the cable latch part is stripped on one of the catches and even the drive vibration seemed to make it wonky. I used a small bit of hot glue to stick the cable on and it fixed it immediately.
April 6, 201016 yr Post up your syslog and someone here will make comment of the possible sources of errors you are seeing on disk 1. Of the cuff, reseating the sata cables is the usual suspect followed by replacing the sata cable on that drive.
April 6, 201016 yr Author I will check the SATA cable and replace it if needed. Here is the syslog from yesterday. I will see if I can get one tonight. Might have better info. syslog.txt
April 6, 201016 yr I will check the SATA cable and replace it if needed. For mine, it wasn't the cable at fault. It was the socket on the drive. There are two small square indentions inside the socket where the cable latch nubs fit when you push in the cable. On mine, the latch nub was only latching on one side. The other side is stripped out. No cable will ever latch on that drive so hot glue was the answer. Some times it's the little things that cause the big problems.
April 6, 201016 yr Good suggestion queeg, though does it hold if the SATA cable is accidentally moved or knocked slightly? Personally I wouldn't leave that for long-term use, though I would use it as a method to get the data off the disk, get a replacement and I'd change the disks asap. So I'm just about good to go. All my drives have been added and everything seems to be working ok except I get errors on disk 1 and my server likes to go to sleep for some reason. I have no clue why it's going to sleep. I cannot wake it except for a full shut off and turn back on. I'm going to see if I have a keyboard connected if a key stroke will wake it. This is not a good solution. I need this thing active 24/7. I'm fine with the spin down of the disks just want it available at all times. About the server going to sleep - I had that problem for awhile. It turned out to be one of two issues: Having duplicate files (accidently self induced) in the same shares but on different drives and/or letting the directory cache script cache all directories. It was causing all of my ram to be consumed and unRAID would go wonky. To fix that I set the directory cache to skip a bunch of deep directories and set it to 3 deep (instead of like 15 or so). And secondly I cleaned up the duplicates which there weren't actually so many but I think it makes unRAID unhappy to have them. About the errors on disk1. I may have mentioned that I have a disk where the cable latch part is stripped on one of the catches and even the drive vibration seemed to make it wonky. I used a small bit of hot glue to stick the cable on and it fixed it immediately.
April 6, 201016 yr Here is the syslog from yesterday. I will see if I can get one tonight. Might have better info. Nothing seriously wrong in that syslog. It does have a larger than normal number of ACPI parsing errors, which *may* indicate it is an early version of its BIOS. I'd keep an eye out for an updated BIOS, but there's generally only a small chance that a BIOS update will help.
April 7, 201016 yr Kaygee makes a good point. I had a bad controller that appeared as drive errors. It could have been the PS but I ruled that out through some creative testing. If you have a MB laying around maybe you can give that a go before buying something. Swiching MBs is easy. As previously explained document the drives layout and have it in front of you for the process. db
April 7, 201016 yr Buggy ACPI in BIOS can bring funny problems. In a few other cases flashing the BIOS with a new version fixed the ACPI errors.
April 7, 201016 yr Correct. The best way to do a board change is to do the following: 1). Go to the Main web gui of your unraid server. 2). Make a hard copy and/or soft copy (ie: screen shot saved as a *.jpg or something saved on a desktop, not your unraid server) and have a print it out in hand. Note: Also you can do this from the Devices link of where you can reassign the disks when the disk array has stopped. 3). Replace the old board, install the new board, try to re-cable and re-connect the disks exactly what you had before (ie: disk1 was connected to sataport1, disk2 was connected to sataport2, disk3 was connected to sataport3, etc...). 4). Power up you server, make the appropriate BIOS changes/setting to your liking and see if all the disks are detected in BIOS and in the BIOS of any additional SATA RAID cards that you have (They should, if not, double-check you cabling and power going to the disks/board). 5). Boot up unraid from the USB drive and if you have a screen connected to it, see if the bzroot and bzimage boot fine and the last thing you'll see on the screen is a login prompt. 6). Now go to the web gui of your server, if some disks appear to be missing (which can happen, don't worry), using the hard copy print out (or soft copy you saved before), reassign the disks in the same order (according to the serial number of the disks on the print out/screen shot) the the way it was installed previously. 7). Once your 110% that the disk order is right, then start the array. By this point you could have a fully functional server again, just like before. Kaygee makes a good point. I had a bad controller that appeared as drive errors. It could have been the PS but I ruled that out through some creative testing. If you have a MB laying around maybe you can give that a go before buying something. Swiching MBs is easy. As previously explained document the drives layout and have it in front of you for the process. db
April 8, 201016 yr Good suggestion queeg, though does it hold if the SATA cable is accidentally moved or knocked slightly? Personally I wouldn't leave that for long-term use, though I would use it as a method to get the data off the disk, get a replacement and I'd change the disks asap. About the errors on disk1. I may have mentioned that I have a disk where the cable latch part is stripped on one of the catches and even the drive vibration seemed to make it wonky. I used a small bit of hot glue to stick the cable on and it fixed it immediately. It's really solid now and it would take a lot more than a slight knock for the hot glue to come loose. It's simple to remove the glue with a screw driver or something that can get under it and peel it up. But I could pretty sure I could dangle the drive from the cable without it coming off or even wiggling.
April 8, 201016 yr Yeah it is good thing to know queeg. I would of used a dab of silicon to do this kind of trick, but it takes a lot longer to dry and albeit not as strong too. Good suggestion queeg, though does it hold if the SATA cable is accidentally moved or knocked slightly? Personally I wouldn't leave that for long-term use, though I would use it as a method to get the data off the disk, get a replacement and I'd change the disks asap. About the errors on disk1. I may have mentioned that I have a disk where the cable latch part is stripped on one of the catches and even the drive vibration seemed to make it wonky. I used a small bit of hot glue to stick the cable on and it fixed it immediately. It's really solid now and it would take a lot more than a slight knock for the hot glue to come loose. It's simple to remove the glue with a screw driver or something that can get under it and peel it up. But I could pretty sure I could dangle the drive from the cable without it coming off or even wiggling.
Archived
This topic is now archived and is closed to further replies.