Q: Time Machine Frequent "Start New Backup"

I get this error regularly, and I'm doing a *wired* backup over gigabit Ethernet.

The Garth Gillespie fix doesn't work for me, either (even with the alternate "if this fails..." fix):

iMac-3:~ root# fsck_hfs -drfy /dev/disk3s2

journal_replay(/dev/disk3s2) returned 0

** /dev/rdisk3s2

          Using cacheBlockSize=32K cacheTotalBlock=81920 cacheSize=2621440K.

   Executing fsck_hfs (version hfs-226.1.1).

** Checking Journaled HFS Plus volume.

** Detected a case-sensitive volume.

   The volume name is Time Machine Backups

** Checking extents overflow file.

** Checking catalog file.

** Rebuilding catalog B-tree.

Extent records for rebuilt file 4:

          [ 19697618, 210 ]

          [ 991841, 208 ]

          [ 19687630, 344 ]

          [ 19697030, 273 ]

          [ 61058765, 295 ]

          [ 8729268, 305 ]

          [ 19689209, 462 ]

          [ 64026391, 84431 ]

hfs_UNswap_BTNode: invalid node height (1)

btree file 4:  1000 records

[additional btree file info omitted]

RebuildBTree - InsertBTreeRecord failed with err 35 0x23

RebuildBTree - numRecords = 103001

RebuildBTree - record 50 in node 2397 is not recoverable.

xxxxxxxx BTreeKey (length 8) xxxxxxxx

06 00 3A CB 00 00 00 00                           ..:.....

xxxxxxxx BTreeData (length 52) xxxxxxxx

04 00 00 00 39 CB 00 00 15 00 54 00 72 00 61 00   ....9.....T.r.a.

6E 00 73 00 6C 00 61 00 74 00 65 00 64 00 53 00   n.s.l.a.t.e.d.S.

74 00 72 00 69 00 6E 00 67 00 73 00 2E 00 74 00   t.r.i.n.g.s...t.

78 00 74 00                                       x.t.

** The volume Time Machine Backups could not be repaired.

          volume type is pure HFS+

          primary MDB is at block 0 0x00

          alternate MDB is at block 0 0x00

          primary VHB is at block 2 0x02

          alternate VHB is at block 5641594414 0x15043e62e

          sector size = 512 0x200

          VolumeObject flags = 0x07

          total sectors for volume = 5641594416 0x15043e630

          total sectors for embedded volume = 0 0x00

          CheckHFS returned 8, fsmodified = 1

May I suggest code42 software.   It allows backups to several different locations at the same time.  The software is free.  If you back up to their vault you pay for the storage.  I do confess that the user interface is confusing.  I back up to their vault and to a local mini.  I'm also using Time Machine to an apple wireless tower.  Between the three options, I seem to be covered.  When TM screws up, it is merely irritating vs disastrous. 

I think I paid $50 for 3 years of their vault, but the prices have since gone up.

https://www.code42.com/crashplan/

I've got word that this is fixed in 10.9.5 (latest Mavericks). My informal testing seems to agree – I haven't been experiencing it although I have deliberately disconnect the network a few times while Time Machine is running.

If you've upgraded to 10.9.5 and still experiencing the same issue, please collect diagnostics data, file a bug report and mention the original bug number: #15560145.

To collect diagnostics, go to Terminal and then run the following command:

sudo tmdiagnose

It will generate a report on the desktop. You'll need to attach this file on your bug report. It would be best if you run this command before you start a new Time Machine backup (e.g. while the "start new backup" prompt is still showing).

Apparently not. I just encountered this again on 10.10.1 and reaised a new bug report:

19067934

No, not fixed for me on 10.95. I've had a new imac + Time capsule (backed up over ethernet) since Jan 2014, and todays "Cannot back up etc..." is the third, so about once every 3 months, regularly. Not good enough obviously; full backup aside I lose the continuity. I haven't had the time, but I'm sick of this and will try calling via Applecare - they've been good in the past. I've got no doubt Yosemite won't be a fix for this.

I'm surprised, as I had two previous TCs running for 5 years under 10.58 without problem, except for one packing up after two years, replaced under warranty. It's interesting to hear the problem is not limited to Time Capsules.

I've missed the boat on the diagnostics this time, but I'll make a note for the next time this happens - no doubt it will.

I just discovered this issue last night.

I'm using a WD drive connected via USB to my MacBook Air with 10.10.1

The 250gb backup finished but somehow corrupted he driver to where it says it needs to be formatted again.

This same thing happened when I attempted to backup via wireless with my Airport Extreme.

The OS/Time Machine def does something to the driver over a period of time that corrupts it.

Same issue.  Garth's fixes didn't do the trick.  WD MyCloud to Yosemite iMac.

Anyone have any ideas?

The issue persists in 10.11 GM. Need a louder and bigger mass of users to file bug reports and complain to Apple about this.

I've reported it - who else has?

does it persist in the official 10.11. version?

above it says bracnhed to a new discussion https://discussions.apple.com/thread/7176964

i cant access this page, anyone got another link?

I have the same issue with only one machine.  All others on my network are unaffected.  The one difference that I notice is the affected machine ofen runs with a mounted sparsebundle image.  I think that time machine may be confused while backing up a mounted volume from an image on the disk.  If I make sure that the time machine runs when the sparsebundle is not mounted, I have gone months without a single problem.

To be clear I trigger the time machine failure when:

1. I have an encrypted sparsebundle "SecureDisk"

2. "SecureDisk" is mounted.

3. I allow time machine to backup while "SecureDisk" is mounted.

Question: Are the other people having this issue using mounted volumes from images located on the disk being backed up?  Maybe this is something in common for some of the people having this issue?

Still have this problem between my new MacBookPro and my WD MyCloud.  Seems to occur when the MacBookPro leaves home and returns.  Totally unacceptable to have to re-backup and entire computer for this reason!  Does not affect wife's MacBookPro at home or the iMac (wired in).

I have had this problem as well. I have to delete my Time Machine backup every few days (sometimes even more frequently). It is very frustrating! I filed a bug report, but Apple said that they do not support my Buffalo NAS, and would not offer any additional support.

Pretty sad that this issue has been going on for YEARS with no solution.  I back up using Time Machine in the Server app to a wired drive. Several machines do this constantly.   But, any machine that has a hard drive directly connected never has this issue.  I thought I'd find a solution here, and I guess I did.... stop using time machine in Server and have a hard drive attached to each Mac for backup.  Clunky, but better than not really having a backup. Because if my backup needs to be redone daily, time machine is utterly pointless.