jowi Posted February 25, 2019 Share Posted February 25, 2019 (edited) I've only been running unraid with 4GB of memory for now, which is enough just for the array, and a handfull of dockers like sabnzbd etc. but i wanted to use VM's as well, so i decided to add some more... which started an avalanche of issues... First the memory i ordered got lost in the mail, but sometimes later i got my 3 extra 4GB DDR3 non-buffered ECC sticks... for a total of 16GB (i tried finding 2x8GB but could not find it...) Then i made a mistake, i thought if the machine is down, i can also update the BIOS of the Supermicro X9SCM so it is spectre protected.... if you change something, never change more then 1 thing at a time... So i put in the 16GB, updated BIOS, and rebooted. It worked! For a day and a half... after creating a VM and playing with it, suddenly unraid froze. Host down. Nothing. Rebooted, was up again, and an hour later... froze again. Rebooted, and then suddenly it got stuck at loading bzroot.... problems got worse. Tried a few times, sometimes it booted, sometimes it couldn't get past bzroot... and if it booted, it ran for 20-30 minutes and froze again. So... i booted into the memtest86, suspecting there was something wrong with either BIOS or memory. I don't know how to downgrade the BIOS, so i ran the memorytest for 5 passes, almost 20 hours, without any issues or errors... so i guess the memory is fine? Rebooted again... now the usb stick never gets past bzroot... no matter how much i rebooted. So i took it out, put it in a windows machine, and... it had errors. Fixed it, put it back in, but no matter what i do, i can't get past bzroot... ?!? It just won't boot anymore... So i got a brand new 32GB usb stick, downloaded the windows usb creator and created a new unraid trial version, just to get it to boot.... put that one in, and... i still can't get past bzroot... so... now i know this is either due to the new BIOS, or... the memory is not ok, even though 20 hours of testing showed no errors. So i took the 3 new sticks out, left the original 4GB stick and... it boots right up... the new usb stick, and the old one... I'm a bit relieved that i can get it to boot again (i immediatly made a backup of the usb stick from the flash menu) but i do think it's weird the memory shows no errors while testing, but once i boot unraid, it gets stuck... what could cause this? Or should i run the memory test much longer? Are there any other tests? The 4 memory sticks are the same, same make and type (Kingston DDR3 1333 4GB non-buffered-ECC) Edited February 25, 2019 by jowi Quote Link to comment
Frank1940 Posted February 26, 2019 Share Posted February 26, 2019 Some MBs are very picky on Memory. The problem is often revealed as you install more 'sticks' of memory into the MB. You should probably check to see if your memory is on your MB manufacturer's list of 'approved' chips for that board. You could also google to see if that particular MB is picky about memory. Quote Link to comment
jowi Posted February 26, 2019 Author Share Posted February 26, 2019 The Kingston memory i'm using is on Supermicro's list of approved memory for this board. I've been running with one 4GB stick for over 5 years now. I've bought 3 more of these same memories last week so i could fill all the slots for a total of 16GB. I've ran the memtest86 on the total of 16GB for over 20 hours, finding no errors... but i couldn't get unraid to boot, until i removed the 3 new sticks and left the original one... it booted right up, no problem... so you would think one or more of the new sticks is bad, but if it is, why didn't memtest find it? Yesterday i've added 1 extra memorystick (so the original 4GB + a new one) for a total of 8GB, to see if it would boot. It did, it's been running for almost 16 hours now without any issues. I will add the rest in a couple of days to see what the effect is... Quote Link to comment
Frank1940 Posted February 26, 2019 Share Posted February 26, 2019 As I understand it, the problem is usually that the line drivers can't maintain the proper waveform so adding one chip usually doesn't trigger the problem. It seems like when you get more than two sticks installed that it rears its head. Quote Link to comment
jowi Posted February 26, 2019 Author Share Posted February 26, 2019 Could be.. but you would think memtest would find errors in that case? Anyway, running with 2x4GB (8GB total) went fine, just put in a 3rd 4GB stick for a total of 12GB. Booting went fine, so far so good. If this keeps working for a couple of days i'll put in the 4th stick for a total of 16GB. Quote Link to comment
jowi Posted March 20, 2019 Author Share Posted March 20, 2019 I'm lost here. Running the X9SCM with 3x 4GB for a total of 12GB works just fine. If i put in the 4th RAM, the BIOS shows 16GB, but UNRAID just hangs on loading bzroot... So thinking maybe the RAM is faulty, i orderder another 4GB DDR, same model, same type. And... nothing. BIOS shows 16GB but UNRAID fails to boot... if i switchs things around, no matter which RAM i put in which slot, once i'm over 12GB UNRAID is stuck... one thing i know, all RAM sticks are ok. No faults. How is that possible? Quote Link to comment
Frank1940 Posted March 20, 2019 Share Posted March 20, 2019 Try putting one stick in at a time. (I would start with the last one if you can remember which it is.) Test all four sticks this way. You might also google this MB and see if it is picky about RAM.. By the way, the version of memtst included with Unraid will not find a problem in ECC memory. I believe there is a version which will but you would have to boot it separately. Quote Link to comment
Hoopster Posted March 20, 2019 Share Posted March 20, 2019 I'm lost here. Running the X9SCM with 3x 4GB for a total of 12GB works just fine. If i put in the 4th RAM, the BIOS shows 16GB, but UNRAID just hangs on loading bzroot... Although this may very well be RAM related and you should keep testing that, often getting stuck at bzroot indicates the system can no longer boot in legacy mode and you should try UEFI boot (or vice versa). This can happen after a BIOS or UnRAID/Linux kernel update as it did with my board. I switched to UEFI boot and all was well. If your boot method is currently legacy, rename the -EFI folder on your flash drive to EFI and make any adjustments in BIOS that may be necessary for UEFI boot and reboot your server. Do the opposite to switch from UEFI to legacy. As previously noted, your board may just be picky about how many RAM DIMMs are installed, but it won’t hurt to try switching boot mode. Sent from my iPhone using Tapatalk Quote Link to comment
Frank1940 Posted March 20, 2019 Share Posted March 20, 2019 One more thing to look at is 'Where in the boot process is it hanging'? You could either describe or take a picture of the monitor screen. ( Both might be ideal. Just make sure that all of the text is readable.) Quote Link to comment
jowi Posted March 20, 2019 Author Share Posted March 20, 2019 1 hour ago, Frank1940 said: One more thing to look at is 'Where in the boot process is it hanging'? You could either describe or take a picture of the monitor screen. ( Both might be ideal. Just make sure that all of the text is readable.) It hangs on the 2nd line when booting: "loading bzroot..." Quote Link to comment
jowi Posted March 20, 2019 Author Share Posted March 20, 2019 (edited) 1 hour ago, Hoopster said: Although this may very well be RAM related and you should keep testing that, often getting stuck at bzroot indicates the system can no longer boot in legacy mode and you should try UEFI boot Will try that! I can't find any UEFI settings other than a 'UEFI recovery mode' in the BIOS manual... will do some googling on UEFI and the Supermicro X9SCM board first. Edited March 20, 2019 by jowi Quote Link to comment
Hoopster Posted March 20, 2019 Share Posted March 20, 2019 2 hours ago, jowi said: Will try that! I can't find any UEFI settings other than a 'UEFI recovery mode' in the BIOS manual... will do some googling on UEFI and the Supermicro X9SCM board first. Your BIOS may not have many settings directly related to to UEFI. You may notice you have a UEFI:{flash drive} boot drive setting and perhaps some CSM (compatibility support mode) settings that reference UEFI and that may be it (and perhaps not even that). If that is the case, make sure to select them. If not, just try renaming in the -EFI folder on the flash drive to EFI and see what happens when it boots. Quote Link to comment
jowi Posted March 21, 2019 Author Share Posted March 21, 2019 I've renamed the UEFI folder on the sub stick, powered down and set the BIOS boot to use the UEFI version of the USB stick, just to see if it would boot at all. And it did. After that, powered down again, added the remaining 4GB stick (for a total of 16GB in stead of 12GB) and rebooted using the UEFI usb stick. And... it works! So the memory i had wasn't faulty at all... question remains, why won't 'normal' UNRAID boot with 16GB memory? Quote Link to comment
jowi Posted March 21, 2019 Author Share Posted March 21, 2019 *update* Well it did look good but... as soon as i start a VM, unraid freezes instantly and i can only escape with a reboot... tried another 4GB module as well, same effect. For some reason this motherboard won't allow me to use over 12GB if i'm using 4GB modules. For now i take my loss and accept that 12GB is the maximum i can use at the moment. If i run into the correct 8GB modules i might try those, but for now... 12GB is it apparently. Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.