Time Machine Always Hangs At "Finishing Backup"

So here's the latest...

The rebuild after the forced power-down and restart completed successfully. I removed the drive that had rebuilt and replaced it with a cold spare ADM. I initiated a repair/rebuild with the new disk, and it has now completed successfully.

At the moment, Time Machine is once again successfully completing backups. Whether this is because I replaced the disk, or simply because of the reboot/rebuild, I don't know. I'll leave things to run and we'll see what happens. I think it's safe to say that if TM runs without hanging for a week, that would be sufficient proof to me that the issue was in fact due to a flakey disk, and was fixed by swapping it out. I should also mention that I've run Disk Utility's "First Aid" against the Time Machine volume, but not the system boot volume. That'll be my next step, if TM starts hanging again.

BDAqua,

I've uncovered even more information. Not only are TM and DU getting stuck, my system hangs on shutdown. This has happened (hang on shutdown) on at least two previous occasions, but I never connected them to TM and DU getting stuck. This evening, I ran sudo shutdown -h now. The desktop froze and I got the little spinner in the middle of the screen, and that was it... system was hung tight, never moved.

I let it sit for 20 minutes before I did a forced power off. On reboot, I held down Command-R, but the system booted as normal without going into Recovery Mode. When the system came up, there were more fun problems.

First, Disk Utility is now starting normally. It reports that the (software) RAID 1 mirror set of twin internal drives that comprise the system volume is degraded, and disk0s2 currently shows "rebuilding". The disk itself isn't marked as failed or bad, and I don't see anything to indicate the disk is bad--other than the fact that it is currently rebuilding. This has happened the last two times the server hung on shutdown.

Second, OpenDirectory isn't running anymore. Apparently, the abnormal shutdown/forced power off corrupted OD. This has also happened the last two times the server hung on shutdown (corrupt OD). On all three occasions, I have been able to recover OD--thankfully--by rsync'ing the appropriate directory from the most recent TM backup.

I'm beginning to suspect your suggestion is correct--it may well be a problem with one of the two mirrored internal hard drives--although, I have to say, if a failing or flakey hard drive in a mirrored RAID array can bork up the system this badly, it kind of defeats the purpose of RAID-1 and doesn't speak too highly of Apple's software RAID implementation.

At the moment, I'm letting the mirror set rebuild (10% in 30 minutes). While I wait for it to complete, I'm going to check and see if TM will complete a backup now (I'm assuming it will work, since DU now works). When the rebuild it complete, I'll try rebooting to Recovery Mode again and hope for a normal shutdown. I'm really curious now to see what DU will report of the Mirror Set and the health of the individual disks.

Any additional comments or suggestions are welcome. I'll post more as I learn more.

I should also add that I did several more hours of searching a reading online today. I discovered a few message threads where people report two of my symptoms: specifically that DU is hanging, and that their system hangs on shutdown. In most of those cases, the problem turned out to be a bad, flaky, or improperly unmounted/ejected external USB drive. However, in my case, I'm no using any USB drives. As seen in one of my earlier posts, however, it is now starting to look like the problem may have something to do with one of the two internal disks in the RAID-1 mirror set that comprises the system volume.

One final post to wrap up this thread.

Although my root problem isn't solved (it appear to be Apple's software RAID), the issue for which I started this thread has been identified: the reason why Time Machine was hanging at "Finishing Backup" was, in my case, due to that fact that my RAID-1 mirror set was degraded. Once the mirror set was rebuilt and the RAID was healthy--and the server was rebooted--Time Machine resumed normal functioning. A few comments/observations:

• The fact that a failed disk in a RAID-1 mirror set causes Time Machine to hang is inexcusable. That's just poor error handling on the part of whomever is responsible for TM.
• The fact that TM doesn't log what its trying to do when it hangs is also inexcusable.
• The fact that repairing the RAID-1 mirror set isn't enough by itself to allow TM to resume normal functioning, and that the server still needs to be rebooted even ever the RAID has been restored to healthy, is also pretty poor.

Thanks to ShadowDust99 and BDAqua.

I'm going to start another thread on this RAID business.

Hi Shadow,

Thank you very much for your response! I hadn't seen that specific article until you posted the link. After reviewing it, unfortunately, most of it isn't applicable:

• Your first backup -- TM has occasionally gotten past the first backup before getting stuck.
• Large changes -- TM sticks regardless of the size or number of changes, even if there have been few or no changes
• Network speed -- TM sticks when backing up local volumes to local volumes when the network is not involved
• Antivirus software -- I'm not using any
• Check your drives -- I'll admit, I haven't done this and I will give it a try. I highly doubt there is a problem with the backup drive, however, because I've tried using brand-new, blank drives, both as a network share and locally attached. I find it hard to believe that a new, blank, freshly formatted disk would need to be repaired.

I'll try checking my drives and will post back here if that has any impact--but I doubt seriously that it will.

Well this is interesting... I'm trying to use Disk Utility to check all of my disks per ShadowDust99's suggestion, and I can't get DU to start. The GUI comes up but it just says "Loading disks..." with a spinning icon, and nothing happens. I tried logging out and logging back on again, and the same thing happens. Just for grins, I ran a manual TM backup and it still hangs at "Finishing Backup".

I wonder if these two problems are somehow related?

Googling on the DU stuck at "Loading disks", I get a LOT of hits with a lot of threads about people complaining about this problem. Aparently, it became a problem clear back at El Capitan, and has continued right up through High Sierra. While there are a lot of people experiencing the problem, there isn't much talk about a root cause. Several people indicate a runaway fsck process, but there is no fsck process running on my machine. Some also claim it may have to do with a USB device that wasn't correctly recognized, but at the moment I have no USB devices attached to the system. Still other people suggest a reboot to clear the problem--but that isn't a fix, that's a temporary workaround.

Still open to other hints or suggestions.

Well, that didn't take long.

48 hours later and the machine has started exhibiting the exact same symptoms: TM and DU both hanging. Using the diskutil appleRAID command, I discovered that the other drive in my mirror set was now listed as "missing/damaged".

Since I know from experience that performing a shutdown at this point will hang and corrupt OD, I'm trying to do the entire disk replacement and RAID set rebuild from the command line. I successfully swapped out the disk with my last remaining spare ADM, and issued diskutil appleRAID disk2 /dev/disk1, where "disk2" is my RAID virtual volume, and /dev/disk1 is the new physical disk. The RAID mirror set is rebuilding as I write this.

I'm starting to think there's either a bug in Apple's software RAID, or some kind of hardware problem, or both, because I find it extremely difficult to believe that two new drives would fail so close together with such extraordinarily light duty usage. My gut is telling me there's something else wrong here, and failed drives in a mirror set is just a symptom. Even so, it is totally unacceptable that a failed drive in a RAID array causes the DU GUI to hang, and TM backups hang altogether. In fact, TM just so happened to kick off a backup just now while the RAID set was rebuilding (around 4% done), and it's hung just like it has so far. Such behavior isn't exactly "high availability" and doesn't speak highly of Apple reliability.

Next hurdle to cross: when the RAID set rebuild is complete and marked as healthy, will DU and TM resume normal functioning, or will I need to reboot first? And if I have to reboot, will the machine shut down, or will it still hang and corrupt my OD database? Details as events warrant...