Time Machine aborting backups intermittently with error -34 since June 17th security update

Thanks. Yes, I did Safe Boot. Same behavior. Checked both the volume being backed up and the local external drive with disk first aid. Both OK. (the boot vol had a couple of warnings, something about a crypto key and a block count that didn't match what it should have been, but was deemed OK.) I didn't fully check the remote volume because when the remote machine reached "checking catalogue file" (or something like that) it said it had two hours to go — it's an old, slow machine! In any case, my wife's computer has been backing up to that remote volume just fine. I examined the last 28 error log entries more carefully. They were a smattering of other errors besides -43 (-34, -36, 5, and -8062). The ones that really worry me are the two error 5's, which were associated with a getxattr and setxattr. In both cases the file referenced had a garbled name — a couple of extra weird characters added after the correct volume name! In all other cases the errors involved copying something from a snapshot file to either null or a file in an InProgress file on the local or remote backup drive. (Roughly half involve the copy to null, the other half to InProgress.)

Now here's the kicker — I realized after I'd been examining the log and writing this reply that I'd accidentally unplugged power and it has been backing up under battery power for roughly the last hour WITHOUT ANY PROBLEM! It has now backed up an additional 16 GB — more than any time since the 17th — and seems to be chugging along just fine. I'm going to let it keep backing up under battery power and see how far it gets before erroring off or running the battery down. My hunch is it's going to do just fine. If it does, I'll try resetting PRAM and the power controller (SPC? I forget the initials. I've only rarely reset it!)

(I did eventually find out about the unix command that interprets an error code. -8062 returned something like "unknown"!) A bit more to add to the above. The backup continued fine until I got down to 30% battery (with roughly 50% of the backup completed) and had to reattach power since I had to do something else and couldn't babysit the computer. Within a couple of minutes the backup errored off, restarting a couple of times until I was able to unplug power once the battery was full awhile later. It again chugged along fine. I then had to go to bed and reattached power. Again, it errored off. But it had only a few gigs left to go, so it had successfully finished by morning. Sometime before the first time I unplugged power I had started streaming the live log to a terminal window, filtering for debug and higher messages from TM/backup demon. When I examined it the next day I noticed something interesting. Shortly after each time I'd reattached power there was an info message that contained "received unmount notification for stable snapshot" (or something like that; the log's in a place I can't easily get to from this account.) Shortly after that there is a notification that the backup was cancelled. I have not very very very carefully examined the log yet, but it looks like this sequence of messages occurred after every time I reattached power and only then. There are many more "Backup cancelled" messages than just after the power attachments. I haven't examined them closely, but at first glance it's after every time some other error (like the file not found or i/o error ones). I see there's another response to this thread. … to be continued.

re resetting SMC. a Yes, I was aware of those steps. I haven't tried it yet. Although I can't really guess how a buggy SMC would cause the snapshot to be unmounted when power is attached, other than reinstalling the whole OS I can't think of anything else to try!

This series of backups (since sometime yesterday, or maybe even before; I forget) is all to an external drive on the remote Mac. But remember, my wife's computer (a Mac mini) has no problem backing up to the same drive, remotely, and I did get the same problems when I tried to backup to a local external drive. I'm going to see if I can figure/find out how to see the SMC's log entries, both to see if they say anything interesting and to pinpoint precisely the times I re-connected power during this. I noted them down, but only to the nearest minute, if even that. When I get time (which won't be for awhile) I'll post an excerpt from the TM log of everything that happened after re-powering.

Well, I tried resetting the SMC. The only difference I can tell is that now backups fail (typically with error -34) both when on ac or on battery! I think that's a step backwards. Unless somebody has a better idea, I think I'll upgrade to Mojave and see if that fixes the problem. (I have a bootable clone of my system on an external drive now, so I should be fairly easily able to revert to High Sierra from that.)

While your answer didn't directly solve my problem, indirectly it did, so I'm giving you credit. Before installing Mojave I needed to free up some disk space — Mojave needed 12.5 GB and I had something like only 11GB free. So I moved my iTunes library to an external drive (no big deal since I use iTunes match.) That freed up about 20G more space. While I was trying to figure out how to actually get Mojave (I never did: there seemed to be a circular chain of web pages, none of which said "click here to download Mojave" or anything similar) I noticed that a backup was going on — a retry of the backup I'd started after resetting the SMC. It had already backed up something like 10G so I just let it go, without ever getting Mojave. 7 hours or so later it finished just fine, and since then it has done several successful hourly backups. I have a hunch what's going on. When I was re-learning how to use CCC I learned that "snapshots" are a feature of APFS. My hunch is that if there isn't enough free space (what "enough" is I haven't a clue!) TM and/or APFS itself can't properly manage snapshots. (Of course I have no idea why the system doesn't warn you of potential backup problems if there isn't enough free space.)Possible evidence for that hunch is that most of the errors I was seeing had to do with copying or deleting (copy to null) snapshots, presumably because they'd already been deleted by something. Why at first it seemed to work fine when on battery power perhaps is because snapshot behavior depends on whether it's on battery or ac. After the successful backups I checked on errors with TM Mechanic — it actually reported something like 13 errors, even though the backups had been fine. Most or all seemed to have something to do with deleting all or some of an InProgress file; I suppose that might be caused by two backup tasks going on at once, which I suppose is a possible bug somewhere.