2

I am using the following layout for two Airport Time Capsules.

time capsule layout

The second augments (wirelessly) the network of the first. The objective is to have backup redundancy, while still not bothering with managing disks and scheduling backups.

I can "Enter Time Machine" and see the history, but it's unclear which history this is. Is it the first Time Machine? Is it the second? Is it a merge of both?

I can also run from the Terminal sudo emacs -nw and view /Volumes/mybackup[1,2]/MyMacBook Pro.sparsebundle/bands, but even then I can't see inside bands (or I'm losing patience before the contents of a large directory listing appear). Neither can I run a simple

sudo ls -l "/Volumes/mybackup1/My MacBook Pro.sparsebundle/bands/"

How do I confirm that a backup is being made on each disk?

I am hoping there is a better answer than to turn one or the other off and do a singleton setup.

  • I've answered this focusing on the function of TimeMachine - if you want to ask a follow on question please @ me and link - I'd be glad to dig up the command needed to mount the sparse bundle if that's what you're looking for or can't get it from diskutil man pages. – bmike Jul 27 at 17:17
  • @bmike Thanks for the comment/answer. I'm a bit behind. I'm not even sure of the difference. I'm expecting that when I sudo ls -l /Volumes/this_or_that I would see a simple directory listing. There is clearly a separate question right there. :) – Calaf Jul 27 at 17:20
  • Think of the bands like virtual blocks on storage. You are looking at maching named buckets into which macOS writes files. Imagine if you burned a DVD - these files are as if you chopped the drive up into 1,000 chunks and named them 2101, 2102, 2103... You're seeing the street address of the house and the filesystem keeps track of which "house" or "block" where file data resides. You need to assemble the blocks into the whole before you can use ls – bmike Jul 27 at 17:24
1

I love having two destinations and the best way to check is to force two backups, rotating between them each. You’ll know if the file system are clean, the metadata correct and be able to see the actual timing needed to back up to each.

tmutil startbackup --rotation --block && tmutil startbackup --rotation --block

You will get timing on the backup and can monitor the status in another shell

tmutil status

An alternate method would be to get a tool designed to parse the backup manifest like BackupLoupe

Focusing on the bands which are a binary representation of the file system seems like a lot of work when you can just mount the file system and look at the files stored. Each backup interval has a plist file showing what happened on that backup interval.

Side answers to side questions:

  • no the format for sparse bundles isn’t changing - when they fail to mount, you are likely going to either lose the data or wait a very long time and find out you can't write data to that destination again. By long time, I mean a week or more in some cases to repair or check a sparse disk on TimeCapsule due to the slow processor and massive IO count to perform filesystem checks of Time Machine hard links.
  • When you have multiple destinations, the default behavior is to try for round robin, so with two connected destinations, odd hours would go to one destination and even hours to the other. So if you never sleep, instead of 24 intervals going to one destination (default case when you have only one drive) - you would get 12 backups to each of the two. If the machine sleeps, it should track roughly odd/even and the only imbalance happens when one of the two isn’t connected, the system prefers the “correct” destination, but will back up sequential intervals to the remaining destination if it’s the only one that mounts when requested.
  • I don't understand. If the default behavior is to round robin, why should I call tmutil startbackup --rotation --block. Also, from man tmutil I don't see that calling the same command twice will address alternating connected TM disks (or which ones if more than two are connected). Could you provide a reference to why you think the duplication is correct? – Calaf Jul 27 at 18:55
  • The first attempt should go to device A and the second to device B, @Calaf. No reference will have your setup or names. If you set up three destinations, you need three block and rotation to hit them all in sequence. It has always worked this way so my only reference is empirical evidence. – bmike Jul 27 at 18:57
  • Let me try asking a separate question on a related issue. A slight overlap is necessary: apple.stackexchange.com/q/365470/7985 – Calaf Jul 27 at 19:03
  • I’ll leave that for others. I fear I have no clue what you are trying to ask, but maybe after a run and some weekend I’ll see where I’m missing your point. – bmike Jul 27 at 19:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .