January 5, 200818 yr Hi All, I have had a problem for some time (since version 3 I think) where my unraid server will reboot if I spin up multiple disks at the same time. I've lived with the problem because I was the only person using the server and I could control the spinup of individual drives by simply accessing them one by one. If I needed to shutdown, I would simply visit each disk through Windows Explorer and confirmed that they were all spun up before running the STOP script and shutting down. Now that I have more clients connecting to the server, I need to get this problem fixed. Here is how I reproduce the problem easily: 1. I set all disks to spin down. 2. I then hit the button to spin up all disks through the web interface. 3. Immediately after I hit the button to spin up all disks, I visit "M:\" which is mapped to a share on the Unraid server. After step 3, the server will eventually reboot and I never get an explorer window to open M:\. I can reproduce this 100% of the time. After the reboot, various things may happen like: a) The array config is reset and I have to rebuild the parity drive. b) I have to recheck the parity drive. c) Nothing happens and it recovers just fine. I tailed the syslog to try and see if I could get any errors. Here it is: ------------------------------------- Tower login: root [Disconnect bypassed -- root login allowed.] Linux 2.6.22.5. root@Tower:~# tail -f /var/log/syslog Jan 5 02:00:42 Tower emhttp[1092]: shcmd (27): /usr/sbin/hdparm -y /dev/hde >/d ev/null Jan 5 02:00:42 Tower emhttp[1092]: shcmd (28): /usr/sbin/hdparm -y /dev/hdf >/d ev/null Jan 5 02:00:42 Tower emhttp[1092]: shcmd (29): /usr/sbin/hdparm -y /dev/hdg >/d ev/null Jan 5 02:00:43 Tower emhttp[1092]: shcmd (30): /usr/sbin/hdparm -y /dev/hdh >/d ev/null Jan 5 02:00:43 Tower emhttp[1092]: shcmd (31): /usr/sbin/hdparm -y /dev/hdi >/d ev/null Jan 5 02:00:43 Tower emhttp[1092]: shcmd (32): /usr/sbin/hdparm -y /dev/hdj >/d ev/null Jan 5 02:00:43 Tower emhttp[1092]: shcmd (33): /usr/sbin/hdparm -y /dev/hdk >/d ev/null Jan 5 02:00:43 Tower emhttp[1092]: shcmd (34): /usr/sbin/hdparm -y /dev/hdl >/d ev/null Jan 5 02:06:16 Tower in.telnetd[1640]: connect from 192.168.1.4 (192.168.1.4) Jan 5 02:06:17 Tower login[1641]: ROOT LOGIN on `pts/0' from `quadxeon' Jan 5 02:06:49 Tower emhttp[1666]: spinup_disk: /dev/hda Jan 5 02:06:49 Tower emhttp[1667]: spinup_disk: /dev/hdb Jan 5 02:06:49 Tower emhttp[1668]: spinup_disk: /dev/hdc Jan 5 02:06:49 Tower emhttp[1669]: spinup_disk: /dev/hdd Jan 5 02:06:49 Tower emhttp[1670]: spinup_disk: /dev/hde Jan 5 02:06:49 Tower emhttp[1671]: spinup_disk: /dev/hdf Jan 5 02:06:49 Tower emhttp[1672]: spinup_disk: /dev/hdg Jan 5 02:06:49 Tower emhttp[1673]: spinup_disk: /dev/hdh Jan 5 02:06:49 Tower emhttp[1674]: spinup_disk: /dev/hdi Jan 5 02:06:49 Tower emhttp[1675]: spinup_disk: /dev/hdj Jan 5 02:06:49 Tower emhttp[1676]: spinup_disk: /dev/hdk Jan 5 02:06:49 Tower emhttp[1677]: spinup_disk: /dev/hdl Connection to host lost. -------------------------------------------- I read in other posts that this may be a power supply issue. I'm currently running 2 power supplies with the Cooler Master dual-PSU adapter. They were both 350 watt, decent quality PSUs. I tried replacing the second one with a much higher quality PSU and it still did not help. My unraid server is based on the original config specified by Tom (intel motherboard, cooler master case, etc.). I have 1GB of ram and a 1GB memory stick. All drives are IDE and I use the promise controllers specified by Tom in his original server config. Does any one have any ideas? Should I replace both power supplies or try getting 1 good one? (like the PC Power & Cooling one in the latest unraid server build specs) I have also tried: a) changing or reseating some of the IDE cables. b) reseating the promise IDE PCI cards. Help!
January 5, 200818 yr Author I tried this again after my computer rebooted. My server was in a state where all drives appeared as "New" and I have to rebuild the parity. I tried a spindown and then spinup of all drives and got this: ---------------------------------------------------- Tower login: root [Disconnect bypassed -- root login allowed.] Linux 2.6.22.5. root@Tower:~# tail -f /var/log/syslog Jan 5 02:46:22 Tower kernel: [ 197.739497] md4: new disk Jan 5 02:46:22 Tower kernel: [ 197.739637] md5: new disk Jan 5 02:46:22 Tower kernel: [ 197.739775] md6: new disk Jan 5 02:46:22 Tower kernel: [ 197.739914] md7: new disk Jan 5 02:46:22 Tower kernel: [ 197.740051] md8: new disk Jan 5 02:46:22 Tower kernel: [ 197.740187] md9: new disk Jan 5 02:46:22 Tower kernel: [ 197.740324] md10: new disk Jan 5 02:46:22 Tower kernel: [ 197.781965] md11: new disk Jan 5 02:46:24 Tower in.telnetd[1299]: connect from 192.168.1.4 (192.168.1.4) Jan 5 02:46:25 Tower login[1300]: ROOT LOGIN on `pts/0' from `quadxeon' Jan 5 02:46:37 Tower emhttp[1313]: spinup_disk: /dev/hda Jan 5 02:46:37 Tower emhttp[1314]: spinup_disk: /dev/hdb Jan 5 02:46:37 Tower emhttp[1315]: spinup_disk: /dev/hdc Jan 5 02:46:37 Tower emhttp[1317]: spinup_disk: /dev/hde Jan 5 02:46:37 Tower emhttp[1318]: spinup_disk: /dev/hdf Jan 5 02:46:37 Tower emhttp[1319]: spinup_disk: /dev/hdg Jan 5 02:46:37 Tower emhttp[1316]: spinup_disk: /dev/hdd Jan 5 02:46:37 Tower emhttp[1320]: spinup_disk: /dev/hdh Jan 5 02:46:37 Tower emhttp[1321]: spinup_disk: /dev/hdi Jan 5 02:46:37 Tower emhttp[1322]: spinup_disk: /dev/hdj Jan 5 02:46:37 Tower emhttp[1323]: spinup_disk: /dev/hdk Jan 5 02:46:37 Tower emhttp[1324]: spinup_disk: /dev/hdl Jan 5 02:46:55 Tower kernel: [ 230.254086] hdb: dma_intr: status=0x7f { DriveR eady DeviceFault SeekComplete DataRequest CorrectedError Index Error } Jan 5 02:46:55 Tower kernel: [ 230.254096] hdb: dma_intr: error=0x7f { DriveSt atusError UncorrectableError SectorIdNotFound TrackZeroNotFound AddrMarkNotFound }, LBAsect=1103831727999, high=65793, low=8355711, sector=480 Jan 5 02:46:55 Tower kernel: [ 230.254113] ide: failed opcode was: unknown Jan 5 02:46:55 Tower kernel: [ 230.256705] hda: DMA disabled Jan 5 02:46:55 Tower kernel: [ 230.256709] hdb: DMA disabled Jan 5 02:46:55 Tower kernel: [ 230.398294] ide0: reset: success Jan 5 02:47:05 Tower kernel: [ 240.661431] hdg: dma_timer_expiry: dma status = = 0x61 Jan 5 02:47:15 Tower kernel: [ 250.645026] hdg: DMA timeout error Jan 5 02:47:15 Tower kernel: [ 250.645037] hdg: dma timeout error: status=0x50 { DriveReady SeekComplete } Jan 5 02:47:15 Tower kernel: [ 250.645041] ide: failed opcode was: unknown Jan 5 02:47:15 Tower kernel: [ 250.645113] hdg: task_in_intr: status=0x51 { Dr iveReady SeekComplete Error } Jan 5 02:47:15 Tower kernel: [ 250.645119] hdg: task_in_intr: error=0x04 { Dri veStatusError } Jan 5 02:47:15 Tower kernel: [ 250.645122] ide: failed opcode was: unknown Jan 5 02:47:15 Tower kernel: [ 250.645266] hdg: task_in_intr: status=0x51 { Dr iveReady SeekComplete Error } Jan 5 02:47:15 Tower kernel: [ 250.645271] hdg: task_in_intr: error=0x04 { Dri veStatusError } Jan 5 02:47:15 Tower kernel: [ 250.645274] ide: failed opcode was: unknown Jan 5 02:47:15 Tower kernel: [ 250.651030] hdg: task_in_intr: status=0x51 { Dr iveReady SeekComplete Error } Jan 5 02:47:15 Tower kernel: [ 250.651038] hdg: task_in_intr: error=0x04 { Dri veStatusError } Jan 5 02:47:15 Tower kernel: [ 250.651041] ide: failed opcode was: unknown Jan 5 02:47:15 Tower kernel: [ 250.651181] hdg: task_in_intr: status=0x51 { Dr iveReady SeekComplete Error } Jan 5 02:47:15 Tower kernel: [ 250.651188] hdg: task_in_intr: error=0x04 { Dri veStatusError } Jan 5 02:47:15 Tower kernel: [ 250.651191] ide: failed opcode was: unknown Jan 5 02:47:15 Tower kernel: [ 250.651198] hdh: DMA disabled Jan 5 02:47:15 Tower kernel: [ 250.651224] pdc202xx_new: Secondary channel res et. Jan 5 02:47:15 Tower kernel: [ 250.744859] ide3: reset: success Jan 5 02:47:15 Tower kernel: [ 250.821931] md0: new disk Jan 5 02:47:15 Tower kernel: [ 250.822347] md1: new disk Jan 5 02:47:15 Tower kernel: [ 250.822785] md2: new disk Jan 5 02:47:15 Tower kernel: [ 250.822935] md3: new disk Jan 5 02:47:15 Tower kernel: [ 250.823086] md4: new disk Jan 5 02:47:15 Tower kernel: [ 250.823466] md5: new disk Jan 5 02:47:15 Tower kernel: [ 250.823614] md6: new disk Jan 5 02:47:15 Tower kernel: [ 250.823777] md7: new disk Jan 5 02:47:15 Tower kernel: [ 250.823997] md8: new disk Jan 5 02:47:15 Tower kernel: [ 250.824156] md9: new disk Jan 5 02:47:15 Tower kernel: [ 250.824303] md10: new disk Jan 5 02:47:15 Tower kernel: [ 250.824442] md11: new disk Jan 5 02:47:15 Tower kernel: [ 250.861085] hdh: drive_cmd: status=0xd0 { Busy } Jan 5 02:47:15 Tower kernel: [ 250.861091] ide: failed opcode was: 0xb0 Jan 5 02:47:15 Tower kernel: [ 250.864752] hdh: status error: status=0xd0 { Bu sy } Jan 5 02:47:15 Tower kernel: [ 250.864758] ide: failed opcode was: 0xea Jan 5 02:47:15 Tower kernel: [ 250.864774] hdh: drive not ready for command Jan 5 02:49:21 Tower kernel: [ 376.348991] md0: new disk Jan 5 02:49:21 Tower kernel: [ 376.349166] md1: new disk Jan 5 02:49:21 Tower kernel: [ 376.349324] md2: new disk Jan 5 02:49:21 Tower kernel: [ 376.349471] md3: new disk Jan 5 02:49:21 Tower kernel: [ 376.349637] md4: new disk Jan 5 02:49:21 Tower kernel: [ 376.349789] md5: new disk Jan 5 02:49:21 Tower kernel: [ 376.349936] md6: new disk Jan 5 02:49:21 Tower kernel: [ 376.349979] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:21 Tower kernel: [ 376.349983] ide: failed opcode was: 0xea Jan 5 02:49:21 Tower kernel: [ 376.350000] hdh: drive not ready for command Jan 5 02:49:21 Tower kernel: [ 376.350003] md7: new disk Jan 5 02:49:21 Tower kernel: [ 376.350242] md8: new disk Jan 5 02:49:21 Tower kernel: [ 376.350418] md9: new disk Jan 5 02:49:21 Tower kernel: [ 376.350589] md10: new disk Jan 5 02:49:21 Tower kernel: [ 376.350740] md11: new disk Jan 5 02:49:21 Tower kernel: [ 376.408390] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:21 Tower kernel: [ 376.408397] ide: failed opcode was: 0xe5 Jan 5 02:49:21 Tower kernel: [ 376.408402] hdh: drive not ready for command Jan 5 02:49:21 Tower kernel: [ 376.468288] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:21 Tower kernel: [ 376.468295] ide: failed opcode was: 0xea Jan 5 02:49:21 Tower kernel: [ 376.468313] hdh: drive not ready for command Jan 5 02:49:23 Tower kernel: [ 378.459391] md0: new disk Jan 5 02:49:23 Tower kernel: [ 378.462177] md1: new disk Jan 5 02:49:23 Tower kernel: [ 378.462352] md2: new disk Jan 5 02:49:23 Tower kernel: [ 378.462496] md3: new disk Jan 5 02:49:23 Tower kernel: [ 378.462632] md4: new disk Jan 5 02:49:23 Tower kernel: [ 378.471798] md5: new disk Jan 5 02:49:23 Tower kernel: [ 378.479726] md6: new disk Jan 5 02:49:23 Tower kernel: [ 378.479768] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:23 Tower kernel: [ 378.479773] ide: failed opcode was: 0xea Jan 5 02:49:23 Tower kernel: [ 378.479788] hdh: drive not ready for command Jan 5 02:49:23 Tower kernel: [ 378.479792] md7: new disk Jan 5 02:49:23 Tower kernel: [ 378.480590] md8: new disk Jan 5 02:49:23 Tower kernel: [ 378.480730] md9: new disk Jan 5 02:49:23 Tower kernel: [ 378.480869] md10: new disk Jan 5 02:49:23 Tower kernel: [ 378.481003] md11: new disk Jan 5 02:49:23 Tower kernel: [ 378.534898] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:23 Tower kernel: [ 378.534906] ide: failed opcode was: 0xe5 Jan 5 02:49:23 Tower kernel: [ 378.534911] hdh: drive not ready for command Jan 5 02:49:23 Tower kernel: [ 378.594794] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:23 Tower kernel: [ 378.594801] ide: failed opcode was: 0xea Jan 5 02:49:23 Tower kernel: [ 378.594818] hdh: drive not ready for command Jan 5 02:49:33 Tower kernel: [ 388.562159] md0: new disk Jan 5 02:49:33 Tower kernel: [ 388.562332] md1: new disk Jan 5 02:49:33 Tower kernel: [ 388.562491] md2: new disk Jan 5 02:49:33 Tower kernel: [ 388.562638] md3: new disk Jan 5 02:49:33 Tower kernel: [ 388.562925] md4: new disk Jan 5 02:49:33 Tower kernel: [ 388.563071] md5: new disk Jan 5 02:49:33 Tower kernel: [ 388.563217] md6: new disk Jan 5 02:49:33 Tower kernel: [ 388.563260] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:33 Tower kernel: [ 388.563264] ide: failed opcode was: 0xea Jan 5 02:49:33 Tower kernel: [ 388.563280] hdh: drive not ready for command Jan 5 02:49:33 Tower kernel: [ 388.563284] md7: new disk Jan 5 02:49:33 Tower kernel: [ 388.563522] md8: new disk Jan 5 02:49:33 Tower kernel: [ 388.563659] md9: new disk Jan 5 02:49:33 Tower kernel: [ 388.563798] md10: new disk Jan 5 02:49:33 Tower kernel: [ 388.563901] md11: new disk Jan 5 02:49:33 Tower kernel: [ 388.618333] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:33 Tower kernel: [ 388.618341] ide: failed opcode was: 0xe5 Jan 5 02:49:33 Tower kernel: [ 388.618346] hdh: drive not ready for command Jan 5 02:49:33 Tower kernel: [ 388.678225] hdh: status error: status=0x58 { Dr iveReady SeekComplete DataRequest } Jan 5 02:49:33 Tower kernel: [ 388.678232] ide: failed opcode was: 0xea Jan 5 02:49:33 Tower kernel: [ 388.678249] hdh: drive not ready for command root@Tower:~# Broadcast message from root (Sat Jan 5 02:50:41 2008): The system is going down for reboot NOW! ------------------------------------------------------------------------------------------------------------------------ Unlike my other attempts, I was still able to view the syslog and I did not get disconnected. Eventually, I just hit the reboot button on the web interface. The difference between this attempt and other attempts was that I changed the power supply connections so that the "better" power supply was connected to the motherboard. It looks like I was having problems with spinning up disk hdh. I noticed that the light on hdh was constantly on and it could not spin up. The web interfaced showed all drives finally spinning up while hdh was still flashing/in a spindown state.
January 5, 200818 yr Author Is there a script on the flash drive that I can alter that will prevent certain drives from spinning down? I think I have identified 3 drives that have problems with spinup and I want to try disabling spin down for only those 3 drives.
January 5, 200818 yr See this thread for some tips on disabling spindown of an individual disk: http://lime-technology.com/forum/index.php?topic=1006.0. I had a Maxtor drive that absolutely refused to spin up, and would go so dead after spinning down that the system had to be hard powered off for more than 10 seconds before it could be detected again. Not even a reboot would wake it up. After the spindown was disabled for the drive, then it would perform adequately (not great, but that was a different issue). As you found after it spun down and you try to awake it, the apparently dead drive causes a lot of errors, then is disabled which messes up your array. You can of course disable spindown globally on the unRAID Settings web page, but I like spindown, so I used the -S0 command (as mentioned in the link above) in the go script to disable it for the Maxtor. I should update that thread, as I have changed my thinking on the most reliable way of getting an hdparm spin command to 'take'. You do not have to disable spindown first, before setting the desired spin down timing. You just have to send the command twice, as it occasionally ignores the first attempt to set it. I don't understand the factors that cause the inconsistency, so I would recommend just sending the spin control command once, and only if it seems unreliable, repeat the command.
January 7, 200818 yr You may have a bad HD that is pulling too much juice when combined with others. Try putting the MoBo on one PSU by itself, and running the drives all on the second (larger) PSU.
January 7, 200818 yr You now have three power supplies and you have tried two of the three combinations ( both originals as well as one original plus the new one). Have you tried the other original in tandem with the new one? This really does seem to be a pure power problem, either with a PSU or perhaps an overly power hungry device (as in a small short in the motherboard or HD). As an aside, I would really like to have support for staggered spin up. I figure we could cut out 100W-200W out of our PSUs if that feature were more widely available. Bill
January 8, 200818 yr Author Thanks for the responses. Yes, I do believe now that this is an issue with my power supply not being able to handle these hard drives. I disabled 3 hard drives from spinning down and I still ran into the problem other hard disks not spinning up. It came from various brands: Seagate, Maxtor (old maxtor before Seagate bought them), and Western Digital. Tom: I am using 1 Enermax 420w power supply and 1 Allied 350w power supply. I realize that the Allied PSU might actually be a piece of crap, but I have run 2 of those in my unraid machine for a couple years now. They seemed ok, but, I think I'm going to get the PC Power & Cooling model. Besides, it will be great to just use one power cord for this server. For now, I have configured 5 drives to never spin down and it seems to be ok. I agree - a staggered power up feature would be peachy.
January 12, 200818 yr Author Hi All, I just wanted to report: I installed a new PC Power & Cooling PSU 610 EPS and all is well now. All drives spin up and I have no problems now! Thanks for everyone who chimed in. Lesson learned: Don't be cheap with Unraid! Quality parts means a great experience! -leonowski
January 13, 200818 yr Author Which PSU was the dog? I actually replaced both PSUs and am now running 1 single PSU. PC Power & Cooling definitely makes the some fo the best PSUs out there. I use them in some of my HTPCs and I've never had any trouble with them. I still don't remember why I went cheap 2 years ago when I first built this machine. Oh yeah, I"M A CHEAP BASTARD! doh! The single PSU didn't have enough molex connectors for all of my IDE drives but it had plenty of SATA power connectors. So, I bought a 4-pack of SATA power to molex adapters. They worked great.
Archived
This topic is now archived and is closed to further replies.