ssb201 Posted March 13, 2017 Share Posted March 13, 2017 I am in the process of building my first UnRAID server and have run into a roadblock. Pertinent Hardware: Dell PERC H200 running IT firmware. HGST Enterprise 8TB SAS drive unknown model expander (server has 12 bays and originally had an LSI 9260-8i controller) I originally set up the array as follows: 960GB Samsung 852T x2 as Cache Pool HGST Enterprise 8TB SATA drive as parity SAS drive above as parity 2x 8TB Seagate Archive drives as array. the page reported the configuration as good, but hitting start would not do anything. I tried changing the number of devices, reformatting the cache drives, and preclearing the data drives. When I then tried to preclear the SAS drive I noticed that the same issue I was having with the array occurred with preclear script. I would start the preclear, but nothing happened. I could find no errors in the system logs. I tried removing the SAS drive and the array started up and built parity. I had previously noticed that the SAS drive also does not have SMART data that all the other drives (including the SATA drives on the same controller) do. I was worried that maybe there was a hardware problem, but I could ssh in- format and mount the drive manually, and it worked. Using the unassigned devices plugin does not work for mounting the drive through the UI. No error, it just does not do anything. I am stumped as to what the issue is. The only apparent variables are that it is a SAS drive and the system gets no SMART data. The kernel will happily mount and use it, so it would appear to be something other than a hardware problem. Any ideas? I can live on single parity for now, but really wanted the protection of dual. Quote Link to comment
JorgeB Posted March 13, 2017 Share Posted March 13, 2017 See if this helps http://lime-technology.com/forum/index.php?topic=48508.msg500313#msg500313 Quote Link to comment
ssb201 Posted March 13, 2017 Author Share Posted March 13, 2017 Hi Johnnie, Thanks for the assist. That setting seemed to help, but not fully. Now when I add the disk to the parity it shows a blue box next to it instead of grey, but it still does nothing when I try to start the array and Parity-Sync. No error, no message, it just does nothing. Quote Link to comment
JorgeB Posted March 13, 2017 Share Posted March 13, 2017 Post your diagnostics after trying to add the disk and starting the array. Quote Link to comment
ssb201 Posted March 13, 2017 Author Share Posted March 13, 2017 Here you go. I spotted nothing other than unrelated SMB issues. tower-diagnostics-20170313-1613.zip Quote Link to comment
JorgeB Posted March 13, 2017 Share Posted March 13, 2017 Problem is almost certainly because of this invalid partition error: Mar 13 16:13:21 Tower root: Creating new GPT entries. Mar 13 16:13:21 Tower root: The operation has completed successfully. Mar 13 16:13:21 Tower kernel: sde: sde1 Mar 13 16:13:21 Tower emhttp: shcmd (587): udevadm settle Mar 13 16:13:21 Tower emhttp: invalid partition(s) Mar 13 16:13:21 Tower emhttp: shcmd (588): rmmod md-mod |& logger Mar 13 16:13:21 Tower kernel: md: unRAID driver removed Can you post the output of: sgdisk -p /dev/sde Quote Link to comment
ssb201 Posted March 14, 2017 Author Share Posted March 14, 2017 (edited) Here you go. I had not looked at the detailed output, but I had compared the description with the other 8TB drive. Now it is true, I did not preclear this drive after changing the display setting. Disk /dev/sde: 1953506646 sectors, 7.3 TiB Logical sector size: 4096 bytes Disk identifier (GUID): 797F2BFA-3E2B-409D-B5A9-BC0C16450B85 Partition table holds up to 128 entries First usable sector is 6, last usable sector is 1953506640 Partitions will be aligned on 64-sector boundaries Total free space is 58 sectors (232.0 KiB) Number Start (sector) End (sector) Size Code Name 1 64 1953506640 7.3 TiB 8300 I tried deleting the partition table using parted and then trying again, it recreates the partition but still does not work. Thanks Edited March 14, 2017 by ssb201 Quote Link to comment
JorgeB Posted March 14, 2017 Share Posted March 14, 2017 Partition looks normal, don't know why it's not accepting it, try this: In case you rebooted in the meantime double check sde is still parity2, if yes first destroy current partition: sgdisk -Z /dev/sde then create a new one: sgdisk -o -a 64 -n 1:64:0 /dev/sde Try to assign it again and starting the array, if still doesn't work grab and post new diags. Quote Link to comment
ssb201 Posted March 14, 2017 Author Share Posted March 14, 2017 No. Same thing happens. tower-diagnostics-20170313-1713.zip Quote Link to comment
JorgeB Posted March 14, 2017 Share Posted March 14, 2017 No more ideas, I posted on the release thread to see if someone from LT can take a look. Quote Link to comment
ssb201 Posted March 14, 2017 Author Share Posted March 14, 2017 Thanks. It really is puzzling. Quote Link to comment
limetech Posted March 14, 2017 Share Posted March 14, 2017 4 hours ago, ssb201 said: Thanks. It really is puzzling. The message "invalid partitions(s)" means that after using 'sgdisk' to write a GPT partition, upon read-back, the exact GPT partition layout, including contents of the protective-MBR is not the same as what was written. Usually this might happen due to disk error, but if that happened there would have been a syslog entry for that as well. Indeed puzzling. There is one odd message in the syslog: Mar 13 14:35:30 Tower kernel: BTRFS: device fsid 78f1f644-dea0-4e1d-a352-abbf2b21afc6 devid 1 transid 10 /dev/sde This happens near start of boot up, when btrfs file system is loading. There is a subsystem called "blkid" that btrfs uses to keep a small in-memory database of the overall system btrfs configuration. Apparently this particular device was once formatted with btrfs as a "whole device", that is, not in a partition (there are no partitions). All I can think might be happening is unRaid tries to write a GPT but then blkid says, "hey this is supposed to be btrfs" so it goes and restores something in the MBR so that it continues to be recognized. This is a rather common "issue" with btrfs (I call it an issue): once a device/partition has been formatted with btrfs it's difficult to purge it off that device/partition. I suggest trying something like this: wipefs -a /dev/sde See if that works. If not, something more drastic: dd if=/dev/zero of=/dev/sde bs=1M count=1 But be very careful with above commands, that "/dev/sde" is indeed the problematic device. 1 Quote Link to comment
ssb201 Posted March 15, 2017 Author Share Posted March 15, 2017 (edited) So I ran wipefs and saw: /dev/sde: 8 bytes were erased at offset 0x00010040 (btrfs): 5f 42 48 52 66 53 5f 4d /dev/sde: 8 bytes were erased at offset 0x00001000 (gpt): 45 46 49 20 50 41 52 54 /dev/sde: 8 bytes were erased at offset 0x74702555000 (gpt): 45 46 49 20 50 41 52 54 /dev/sde: 2 bytes were erased at offset 0x000001fe (PMBR): 55 aa /dev/sde: calling ioctl to re-read partition table: Success I then tried to add the drive as my second parity drive and still does not work. syslog still shows: Mar 15 13:32:23 Tower emhttp: shcmd (4003): sgdisk -Z /dev/sde &> /dev/null Mar 15 13:32:23 Tower kernel: sde: sde1 Mar 15 13:32:24 Tower emhttp: shcmd (4004): sgdisk -o -a 64 -n 1:64:0 /dev/sde |& logger Mar 15 13:32:25 Tower root: Creating new GPT entries. Mar 15 13:32:25 Tower root: The operation has completed successfully. Mar 15 13:32:25 Tower kernel: sde: sde1 Mar 15 13:32:25 Tower emhttp: shcmd (4005): udevadm settle Mar 15 13:32:25 Tower emhttp: invalid partition(s) Mar 15 13:32:25 Tower emhttp: shcmd (4006): rmmod md-mod |& logger Mar 15 13:32:25 Tower kernel: md: unRAID driver removed Edited March 15, 2017 by ssb201 Quote Link to comment
JorgeB Posted March 15, 2017 Share Posted March 15, 2017 Try dd On 14/03/2017 at 5:05 AM, limetech said: See if that works. If not, something more drastic: dd if=/dev/zero of=/dev/sde bs=1M count=1 But be very careful with above commands, that "/dev/sde" is indeed the problematic device. Quote Link to comment
JorgeB Posted March 15, 2017 Share Posted March 15, 2017 You could try preclearing the disk but doubt that will help since all the partition info is on the first few sectors, maybe Tom has other suggestions. Quote Link to comment
ssb201 Posted March 16, 2017 Author Share Posted March 16, 2017 (edited) I tried preclearing without any success. I am hoping this is a bug that the lime-tech guys can get to the bottom of. The only other option I can see is using snapraid to get semi-equivalent parity protection using the drive. Edited March 16, 2017 by ssb201 Quote Link to comment
JorgeB Posted March 17, 2017 Share Posted March 17, 2017 (edited) Can you post the output of: dd if=/dev/sde bs=2048 count=1 | hexdump -C Assuming disk is still sde and after trying to add it to the array. Edited March 17, 2017 by johnnie.black Quote Link to comment
xodusgenesis Posted March 17, 2017 Share Posted March 17, 2017 I'm also having the same exact issue. New 8tb drive assigned to parity will keep giving the "invalid partition" error. Successfully precleared the drive too. Had to go back to my current array of 4tb drives in order to use my system. Quote Link to comment
ssb201 Posted March 17, 2017 Author Share Posted March 17, 2017 The output from the hexdump: root@Tower:/# dd if=/dev/sde bs=2048 count=1 | hexdump -C 00000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| 1+0 records in 1+0 records out * 000001c0 02 00 ee ff ff ff 01 00 00 00 55 25 70 74 00 00 |..........U%pt..| 2048 bytes (2.0 kB, 2.0 KiB) copied, 0.000467479 s, 4.4 MB/s000001d0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| * 000001f0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 55 aa |..............U.| 00000200 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 |................| * 00000800 Quote Link to comment
JorgeB Posted March 17, 2017 Share Posted March 17, 2017 (edited) @limetech Just notice this, pretty sure this is the problem: Mar 13 14:35:30 Tower kernel: sd 1:0:1:0: [sde] 1953506646 4096-byte logical blocks: (8.00 TB/7.28 TiB) I believe 4Kn disks are currently not supported by unRAID. There is a similar 512e model ( HUH728080AL5200 instead of HUH728080AL4200) Edited March 17, 2017 by johnnie.black Quote Link to comment
JorgeB Posted March 17, 2017 Share Posted March 17, 2017 (edited) 36 minutes ago, xodusgenesis said: I'm also having the same exact issue Disk model or syslog? is it 4Kn also? Edited March 17, 2017 by johnnie.black Quote Link to comment
ssb201 Posted March 17, 2017 Author Share Posted March 17, 2017 That would explain it. I hope support comes soon. I do not have another 8TB drive hanging around at the moment. Quote Link to comment
limetech Posted March 17, 2017 Share Posted March 17, 2017 Nice find johnnie.black! Damn I should have seen that... That particular model HUH728080AL4200 is indeed a 4K-logical-sector size device: https://www.hgst.com/sites/default/files/resources/Ultrastar-He8-DS.pdf ...and right, unRaid OS does not properly handle that at present. Quote Link to comment
xodusgenesis Posted March 20, 2017 Share Posted March 20, 2017 When are 4Kn drives expected to be supported? I have the same 8tb drives but as SATA. 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.