On December 23, the boot drive in my MBP failed and I had to interweave backups and recovery and troubleshooting through the additional activities of the holidays. As I usually do when experiencing hardware issues, I journaled it. However, for reasons unknown even to me, I kept the journal on Facebook rather than here.
Today, with the removal of the failing drive and its’ replacement with a 1TB SSD, I am done with the portion of tasks associated with resolving this issue that have to do with physically troubleshooting and replacing hard drives. There is a longer-term set of tasks upcoming involving the machine-by-machine deployment of outboard backup volumes, signaling the definitive end of my five years of desultory seeking to create a LAN-based backup system, but that’s a series of blog posts for another day.
Here’s all the stuff I posted on Facebook regarding this issue from December 23, 2016 until today, January 3, 2017. I have removed posts that were from friends and observers here; they can still be seen on the original FB thread.
I was awakened this morning by my parents urgently calling to ask if I could locate some financial information, one of their HDs having failed, I think, although it was somewhat hard to follow their explanation of the issue.
December 23, 2016 at 12:00pm
I was able to mount the MPB as an external drive and am now ‘imaging’* the data over to another outboard. Roughly 600gb to sling, running at 16gb/hr give or take. The process ought to complete in 37 hours or so. Aargh.I cloned a 1TB SSD to a 2TB last night, coincidentally. A mere 12 hour process with today’s blazing transfer speeds. At least we don’t have to sit with the damn machine rotating floppies or swapping DVDs out any more.Oh, wait, hold on: SuperDuper just choked! Damn data integrity shinola!
*not actual, you know, bitwise imaging
December 23, 2016 at 12:57pm
hm, might be well fucked. I’ll look at CCC after I run an errand or two. I moved to SD! after CCC for some reason I now forget.
After SD! choked I went back to recovery mode and told DU to image the bad volume to the external and that failed too with an i/o error. SD! gave me a diagnostic on an individual file and halted instead of skipping and moving forward with a log, which would have been my preferred option.
I think my next best options here are directory-tree restricted copies and after that manual copies. I don’t actually think there’s anything unique or mission critical on the machine that’s not already mirrored, I just didn’t want to have to do a hand rummage and compare if I didn’t have to.
I set up the drive after the last machine failed this summer days before a revenue gig and I had to source this axe right quicklike. That machine was still on Mavericks and I migrated everything to Yosemite after I wrapped that gig. The old Mavericks boot drive is still viable and in the old machine, and I know it can reboot this machine, so I do have a known-good outboard boot solution.
All this fucking travel this year has prevented me committing the week or so of sustained effort to get that last-mile stuff set up. Bummer. At least the local mirroring stuff seems bulletproof, it’s twice this year that’s saved my bacon.
Well, disconcertingly, the Pro just self-booted in the middle of various tasks including running a CCC copy. CCC was able to pick up and move forward, though. CCC also reports physical read errors and, thankfully, does NOT abort the copy.
December 24, 2016 at 1:03am · Edited
Well, good news, and although I can’t match CCC’s bad file report with the size of the uncopied data, there’s nothing here that should prevent the cloned drive from booting. Trying tomorrow. Still probably best to wipe and bump to Sierra and then migrate. Man, I HATE running the current version of an OS release, it just offends my sensibilities. Maybe I can rig a work around to run that Yosemite clean install again. There’s nothing that makes me happier than defeating installer restrictions.
December 24, 2016 at 11:19am
Cloned drive boots. Now to ponder whether ’tis nobler in the eyes of man to clone and clone again, to a drive week-wiped free of foul corruptions’ stink, or to take up arms, and by opposing the screwdriver’s lance to the helices of steel that bind the shield, expose the innermost parts of that foe and servant, thereby to effect a transplant?
December 24, 2016 at 11:27am · Edited
I had actually planned on a drive swap to an SSD on this machine after Xmas. I still need the machine to work on a couple things before then. Opening and swapping drives two times in rapid succession increases the chances of zapping the mobo. Recloning it it is, because what s a day between friends?
December 24, 2016 at 11:55am
hm, I just had a thought – I think my old-style cloning of bootable volumes as a backup and migration strategy is probably not viable longterm any more because it won’t migrate the recovery partition. so even when I get this back in working shape I will need to do an install/migrate route for the SSD later on.
December 24, 2016 at 3:58pm
CCC has a Recovery HD creation function, cool. But it requires an extant recovery volume, and I have just noticed that my current primary boot vols were all stood up as clones, and thus lack the partition.
I will direct a squinty-eyed gaze at the one volume that did have said partition, the MPB drive that failed, kicking off this tech struggle journal.
Normally I keep a technical journal of these incidents on ye olde blogge, and these will migrate there in time.
I think the thesis is that I may record my mistakes, and learn from them.
December 28, 2016 at 12:42pm
Status update: internal drive reformatted and restored from successful CCC backup clone. No recovery partition in place yet. That requires a clean install of Yosemite to some machine or other, something that will require another hour or so of supervision and drive swapping, then CCC can clone those volumes as needed.Next steps: transfer this set of tech logs to the blog, finish deferred production tasks on the MPB, begin the SSD migration on the Pro, finalize SSD install to this machine. I’ll give myself a calendar week for each of these.
December 30, 2016 at 12:34pm
In working toward getting a clean install of Yosemite on an outboard drive so that CCC can harvest and distribute the recovery partition I ran afoul of a driver conflict on the Mac Pro – that machine has had an NVIDIA GTX 970 as its primary video card for a couple of years now, and running late-model NVIDIA cards in post OS X 10.7 systems requires custom drivers, without which the machine goes into a boot-looped kernel panic. Naturally, the temporary installer volumes created when you run an OS X installer lack this driver, and boom, a day of wondering where the heck the right web forum was again.Anyway, clean volume created. Next challenge is migrating the 1TB SSD to the 2TB SSD and re-merging the 2TB vol with the old Mavericks 2TB vol document directory that went from primary to offline when I moved to the SSD.
I guess I could hand merge, since the local Yosemite docs dir is intentionally thin and all the applications and licenses are on the SSD. What really want is a tiered local torrent cloud sync, where the current solution I use mixes with a longer-term storage solution for older files and docs. just figuring out the ruleset is time consuming enough that I doubt I’d ever try to set something like that up though.
December 30, 2016 at 4:21pm
Alrighty, both SSDs boot and have recovery partitions on ’em. Now to research folder-based one-off sync tools. Seems like CCC or Super Duper should just have that as a core feetch.
January 3, 2017 at 4pm
Closing in on winding this up. The initial drive failure appears to have recurred on the same drive, so it’s likely an unrecoverable hardware fault, although I haven’t traced that down to the bare iron yet. I was anticipating this as a possibility and had prepped to swap in the 1TB SSD in that event, which has now been done.
The last little cleanup bits are re-merging the various disaporan Documents folders, something that, as hoped, CCC can handle, but which will require a written plan and a couple of afternoons. Once that’s been done, I need to distribute local backup drives to all the machines and get that set up and I should be hard-drive circus complete for another year or more. I fucking hope.