Time Machine: "Partially Deleted Backup"

Question

Last night, after Time Machine performed a backup and began its post-backup thinning, it got stalled on "Finishing backup... ." The system log showed that it was attempting to delete a previously partially deleted backup:
Starting standard backup
Starting post-backup thinning
Found partially deleted backup - trying again to delete: 2009-09-30-110803
The backup it was trying to delete was the last one on its list (i.e., the oldest one on that TM volume). And when I opened it up, it indeed appeared to be a partially deleted folder. So I let TM run. However, it never finished "Finishing backup... ," so after letting it run all night and all day, I simply told TM to stop. It did, and the system then added two more messages to the log, acknowledging my cancelation as well as TM's current success:
Starting standard backup
Starting post-backup thinning
Found partially deleted backup - trying again to delete: 2009-09-30-110803
Backup deletion was canceled by user
Backup completed successfully.
However, the next time TM ran, it began all over again:
Starting standard backup
Starting post-backup thinning
Found partially deleted backup - trying again to delete: 2009-09-30-110803
It's still sitting there in its "Finishing backup..." mode.
I was thinking about entering Time Machine, selecting that backup, and telling Time Machine to delete that backup—and only that backup. But I'm (a) not sure that will address the actual problem, and (b) wondering if deleting that particular backup (i.e., the oldest one on the list) is advisable.
Suggestions?

Answer 1

Pondini wrote:
Maxwell’s Demon wrote:
I'm aware of the multilink nature of TM's backups. What I don't understand is how you immediately concluded that the backups were corrupted.
TM finding partially deleted backups, trying repeatedly to deleted them, and failing to.
(And yes, I though you meant that you'd tried to delete the backup yourself.)
It took forever. I ran TechTool Pro, and it "choked" (i.e., ran out of memory) trying to rebuild the directory. When I asked Disk Warrior to graph the directory, it revealed that it was more than 40% fragmented!
I'm not sure fragmentation is a problem, or even applies, given the structure of TM backups.
So I ran Disk Warrior. It ran without incident. When I then looked inside the TM drive, I discovered that along with rebuilding the directory—which is what I understood its job to be—DW apparently removed/deleted the offending (partially deleted) backup. The next time Time Machine ran, everything went smoothly. It's been running fine ever since.
Great! And yes, there are times it can repair/rebuild TM backups that nothing else can (I guess that's why it costs $100!).
I'm obviously quite pleased with the results: Not only did DW perform an incredible directory rebuild (fragmentation went from more than 40% to less than 1% !!) but it cleared out the partially deleted backup that was causing TM to stumble. But I remain awfully curious: Had I simply gone into Time Machine and deleted the partially deleted backup myself, wouldn't that have fixed the problem? (Granted, DW not only got TM back on track, but it also rebuilt its directory, which is a definite plus in my mind...I'm just trying to understand.)
It's possible you couldn't have deleted it completely, any more than TM was able to. Remember, a single backup folder has hard links to hundreds, thousands, or hundreds of thousands of other items. Just one of those being screwy can have a ripple effect through the whole structure.
And if you could have, it might well have left some of the actual backup files "abandoned" -- some or all of the hard links to them deleted, but still somewhere in the disk directory, so still taking up space.
I got the idea from
http://www.macosxhints.com/article.php?story=20090515063602219
which provided a link to the "solution" when partial backups remain that result in TM "error: 11" failures:
http://www.bytebot.net/blog/archives/2008/08/13/fixing-time-machine-backup-faile d-with-error-11
Admittedly, mine wasn't an "error: 11" problem, but it sure seemed similar. My hesitation/concern was due to the fact that the partially deleted backup was the last one (i.e., the "oldest") in my TM volume: I wasn't sure if deleting it would screw up all of the links used by the remaining backups. However, since DW removed it without causing any problems, it seems like I may have been able to accomplish the same result by removing it "manually" from within TM. If that is the case, then it suggests that there may not have been any actual "corruption."
Again, I am very happy with the result, and likely will resort to DW again the next time I have similar problems. The only reason I'm harping on it is simply to try to gain a deeper understanding of how to determine when such problems are indeed due to corruption as opposed to simply an errant file that causes TM to stumble. (Had TM indicated that it stumbled and was unable to successfully perform the backup, then I would have little doubt that there was actual corruption.)