Looks like no one’s replied in a while. To start the conversation again, simply ask a new question.

Time Machine constantly reindexing fresh Time Capsule backup

A lot of people have been reporting that TM seems to be constantly reindexing their backup. I've noticed the same problem. First, it'll often indicate that it is "Indexing Backup..." before backing up, and then for HOURS after it indicates that it's finished, there sparsebundle is still mounted, and it's still running mds on it. It does this at least once per day.


I've looked at the various solutions, but they're not helpful to me. I'm hoping there might be more suggestions.


First of all, this is a fresh backup. I created a new one when I installed Lion, so wiping it out and making a new one would be redundant.


I tried deleting the metadata (using mdutil) so that it would create a fresh index, but that didn't change anything.


I've also tried verfying the backup. You can hold Option while clicking the TM menu item and select "Verify Backups". The icon spun for a few seconds then stopped. Now, when I go back to the menu, "Verify Backups" is grayed out. There is no background fsck process running. It's just decided to disable the option.


Is there anything else I can do?


Lion has already killed battery life, and this is only making things worse.


Thanks!

Posted on Aug 14, 2011 8:16 AM

Reply
Question marked as Best reply

Posted on Aug 14, 2011 8:32 AM

theosib wrote:


A lot of people have been reporting that TM seems to be constantly reindexing their backup. I've noticed the same problem. First, it'll often indicate that it is "Indexing Backup..." before backing up, and then for HOURS after it indicates that it's finished, there sparsebundle is still mounted, and it's still running mds on it. It does this at least once per day.



To test, try (if you haven't already) to connect via an ethernet cable to speed things up.

Then, run TM

Open Applications->Utilities->Console

In the Console Search Box enter "backupd"

If you see "Waiting for Spotlight to finish indexing /Volumes/Time Machine Backups/Backups.backupdb", then wait for the TM Volume to be ejected.

I agree that with a fresh TM Volume that it should not take long, but is you see the above message repeating in Console, then try waitng (could take hours)


Tony

21 replies
Question marked as Best reply

Aug 14, 2011 8:32 AM in response to theosib

theosib wrote:


A lot of people have been reporting that TM seems to be constantly reindexing their backup. I've noticed the same problem. First, it'll often indicate that it is "Indexing Backup..." before backing up, and then for HOURS after it indicates that it's finished, there sparsebundle is still mounted, and it's still running mds on it. It does this at least once per day.



To test, try (if you haven't already) to connect via an ethernet cable to speed things up.

Then, run TM

Open Applications->Utilities->Console

In the Console Search Box enter "backupd"

If you see "Waiting for Spotlight to finish indexing /Volumes/Time Machine Backups/Backups.backupdb", then wait for the TM Volume to be ejected.

I agree that with a fresh TM Volume that it should not take long, but is you see the above message repeating in Console, then try waitng (could take hours)


Tony

Aug 14, 2011 8:58 AM in response to Tony T1

These appear relevant:


8/14/11 11:08:40.161 AM com.apple.backupd: Running backup verification

8/14/11 11:08:40.403 AM com.apple.backupd: Stopping backup to allow backup destination disk to be unmounted or ejected.


I decided to tell it to backup manually, and Console said something about the volume already being mounted. So it seems to be unable or unwilling to unmount the backup.


iStat does not list the sparsebundle as being mounted, but "mount" on the command line does. "lsof | grep Backups" lists mds as having tons of files open in the sparsebundle, like this:


mds 646 root txt REG 14,13 3277 3537709 /Volumes/Time Machine Backups/.Spotlight-V100/Store-V2/9990B315-2B02-49A1-B831-3086FCB187A6/0.indexGr oups


After it backed up, I selected "Verify Backups" again, and I got this on the console:


8/14/11 11:54:21.251 AM com.apple.backupd: Backup verification requested by user.

8/14/11 11:54:21.263 AM com.apple.backupd: Network destination already mounted at: /Volumes/NetworkBackup

8/14/11 11:54:21.698 AM com.apple.backupd: Running backup verification

8/14/11 11:54:21.771 AM com.apple.backupd: Stopping backup to allow backup destination disk to be unmounted or ejected.


Now, "mount" shows that the network AFP volume containing the sparsebundle is stealth-mounted:


afp_0YqqZo1fzLUK00m4oF0Pyetl-1.2e000095 on /Volumes/NetworkBackup (afpfs, nobrowse)


But the sparsebundle is not. Still, there's no fsck going on.


The Time Machie Backups appears in Disk Utility, so I'm going to manually run a verify to see what happens.

Aug 14, 2011 9:44 AM in response to theosib

I decided to tell it to backup manually, and Console said something about the volume already being mounted. So it seems to be unable or unwilling to unmount the backup.


From Terminal enter ls -l /Volumes

2 Volumes are mounted:


Time Machine - MacBook Air (...with your named vol)

Time Machine Backups


If you eject the Vol from the Desktop, only "Time Machine Backups" is ejected. To unmount the other TM Vol, from Terminal enter: diskutil unmount /Volumes/<your TM Vol>



After it backed up, I selected "Verify Backups" again, and I got this on the console:


That verifies the files on your Mac, not the TM Backup


The Time Machie Backups appears in Disk Utility, so I'm going to manually run a verify to see what happens.


Not sure which vol your verifying, but you need to open the sparse bundle and run the verify on "Time Machine Backups". (I think this is what you meant, just checking). Also, might as well have run repair.



iStat does not list the sparsebundle as being mounted, but "mount" on the command line does. "lsof | grep Backups" lists mds as having tons of files open in the sparsebundle, like this:


mds 646 root txt REG 14,13 3277 3537709 /Volumes/Time Machine Backups/.Spotlight-V100/Store-V2/9990B315-2B02-49A1-B831-3086FCB187A6/0.indexGr oups



Looks like Spotlight still needs to index.

Try an ordinary TM Backup (via ethernet cable to speed things up), and wait a few hours. I had to wait 6+ hours, others have reported longer.


Tony

Aug 14, 2011 11:07 AM in response to Tony T1

As far as I know, this never unmounts:


afp_0YqqZo1fzLUK00m4oF0Pyetl-1.2e000095 on /Volumes/NetworkBackup (afpfs, nobrowse)


It doesn't show up in Finder or iStat. Nothing seems to be using it either. The sparsebundle is not mounted either.


Using lsof, I see things like this:


mds 646 root 19r DIR 46,149 534 2 /Volumes/NetworkBackup

diskimage 1356 root 4u REG 46,149 8388608 136966 /Volumes/NetworkBackup/Tomahawk.sparsebundle/bands/1f4

diskimage 1356 root 5u REG 46,149 0 59764 /Volumes/NetworkBackup/Tomahawk.sparsebundle/token

diskimage 1356 root 6r DIR 46,149 264 59760 /Volumes/NetworkBackup/Tomahawk.sparsebundle

diskimage 1356 root 7r DIR 46,149 861244 59763 /Volumes/NetworkBackup/Tomahawk.sparsebundle/bands

diskimage 1356 root 8u REG 46,149 8388608 74827 /Volumes/NetworkBackup/Tomahawk.sparsebundle/bands/1b8


So the backup volume is not mounted, but something called 'diskimage' is accessing its bands and other files.


It's as though mds is trying to index the network volume, but I thought network volumes were not allowed to have metadata... in any case, mdutil tells me that it's not allowed to have any. So what is mds doing indexing a hidden mount point?

Aug 14, 2011 12:59 PM in response to theosib

theosib wrote:


Tony T1 wrote:


I would suggest plugging in the Mac, reboot the Mac, let TM start normally and let it run overnight.

Check the progress via Console (enter backupd in the Console Search Box).


Actually, I've done that a few times already.


Sorry, I thought you said that the TM Volume never ejected automatically and that you had to manually eject.


Tony

Aug 15, 2011 5:40 AM in response to theosib

I'm having the same problem, sadly there is nothing I can add to this so far.


I have tried everything. I deleted the entire Backup Server-Side, set it up from scratch, I waited over night, etc etc...


What seems to happen for me is that once it has backuped it all and had enough time to Index it is indeed done and subsequent updates work as they should.


HOWEVER. If I interrupt TimeMachine (even after the first long index run) either by saying Stop Backing Up or by simply closing my laptop it will reindex the entire thing over WiFi which will again take close to 10 hours. Any time I interrupt it while doing that it seems to forget all progress and do it from scratch.


So if I close the laptop again after having it run for 9 hours I have to go through the 10 hours again.


This is obviously very painful and I have since switched to using an external harddrive as a Time Machine Destination until these obvious issues are fixed.


Sadly I have not been able to find any workaround whatsoever, besides using an external drive. On this drive Time Machine works perfectly, even when I interrupt it.


The server I backup to uses netatalk 2.2.0, a program which has been reported by many others to work, so I guess it's not at fault.


Theosib, are you using netatalk too or a Time Capsule?

Time Machine constantly reindexing fresh Time Capsule backup

Welcome to Apple Support Community
A forum where Apple customers help each other with their products. Get started with your Apple ID.