April 14, 201214 yr 4.7, Rajahal's 20-drive beast tower build (incl. cache drive), and no problems for a good while now - but as of this post-outage restart, it took forever for most of my drives to mount, and one disk (#9) is now red-dotted (but still totally accessible, as the complete array is now spun up in spite of the warning)... My gut tells me it's time to replace that (barely six-month-old, 3x-precleared) drive, but I figured I'd check with the experts here before shelling out ~$130 on another 2TB just to tread water (was just about to start working towards a 5.x 3TB-grade box...) Syslog attached, and any guidance would be greatly appreciated. Thanks in advance! *Clarification - red dot next to Disk #9, orange dot next to "started" (which based on research suggests the disk has failed and needs replacement... so why can I still access it through raw //tower/disk9?)
April 14, 201214 yr Author OK, I've checked cables, running (what seems to be a neverending) "short" SMART test via unmenu, and I'm about ready to run by Best Buy and overpay for a replacement... Can anyone advise any other "last ditch steps" before I start preclearing the new drive? Should I stop the array, remove "bad" disk 9, put "new" disk 9 in its place, preclear it a couple of times, then restart the array so it automatically rebuilds - or is there a different set of steps I should take? Thanks for your patience and help!
April 14, 201214 yr OK, I've checked cables, running (what seems to be a neverending) "short" SMART test via unmenu, and I'm about ready to run by Best Buy and overpay for a replacement... Can anyone advise any other "last ditch steps" before I start preclearing the new drive? Should I stop the array, remove "bad" disk 9, put "new" disk 9 in its place, preclear it a couple of times, then restart the array so it automatically rebuilds - or is there a different set of steps I should take? Thanks for your patience and help! The short test NEVER returns anything. You must simply wait the appropriate time, and then get a new status report. At the bottom will be the results of the short test. A "write" failure will result in the red indicator. It will never go away on its own, even if it was just a loose cable that caused the failure. If the disk is writable, it can be re-constructed and put back in service. There are instructions in the wiki for how to get a disk to be recognized as its own replacement. If you can get a status report out of the drive using smartctl, odds are you'll be able to re-construct it. Regardless, if it is not writable, the reconstruction will not occur, and you can then pay full retail for the replacement disk. Joe L.
April 14, 201214 yr Author Thank you so much for the help, Joe! As soon as I checked the smart status, my fears were confirmed: SMART overall-health self-assessment test result: FAILED! Drive failure expected in less than 24 hours. SAVE ALL DATA. See vendor-specific Attribute list for failed Attributes. So here is my plan, based on what I've read across multiple posts (unless someone tells me I'm doing this wrong - I'll probably hold off until tonight just in case a "STOP BEFORE YOU SCREW SOMETHING UP!" post comes through...) (1) Stop Array (2) Remove dying drive (3) Insert new drive (hot-swap cage) (4) Telnet to tower, run preclear x2 (normally x3, but I hate to leave the disabled disk status for 3 days...) (5) Start array (which automatically starts data rebuild, right?) (6) Nightmare resolved
April 17, 201214 yr Author Fully reconstructed. I'm reminded of the Popeye's Chicken scene from "Little Nicky"... unRAID is f*'n awesome! Thank you again, Joe.
Archived
This topic is now archived and is closed to further replies.