Mac host file system corruption during shrinking of PD13 Win10 VM

Discussion in 'Installation and Configuration' started by Robert Swanson, May 3, 2018.

  1. Robert Swanson

    Robert Swanson Bit Poster

    Messages:
    39
    Greetings,
    My iMac desktop host machine is one month old.
    It has the latest OS version and is also using a 1TB SSD using APFS (Apple File System).
    On the SSD is my PD13 Win10 VM.
    This has happened TWICE now:
    1) starting to reclaiming disk space/shrinking the VM
    2) going on with other business when the ENTIRE machine locks up and becomes unresponsive
    3) I have to power off/on resulting in a corrupted VM file AND a corrupted Mac file system
    4) Disk Utility cannot repair it. The output from the repair attempt from Disk Utility is always:
    Verifying file system.
    Volume could not be unmounted. ==> I also tried repairing it from the boot disk; same results.
    Using live mode.
    Performing fsck_apfs -n -l -x /dev/rdisk1s1
    Checking volume.
    Checking the container superblock.
    Checking the EFI jumpstart record.
    Checking the space manager.
    Checking the object map.
    Checking the APFS volume superblock.
    Checking the object map.
    Checking the fsroot tree.
    error: invalid dstream.size (494593892352), is greater than dstream.alloced_size (391420366848)
    error: xf : INO_EXT_TYPE_DSTREAM : invalid dstream
    error: inode_val: object (oid 0x18471a): invalid xfields
    fsroot tree is invalid.
    The volume /dev/rdisk1s1 could not be verified completely.

    File system check exit code is 0.
    Restoring the original state found as mounted.
    Operation successful.
    5) The only solution is to REINSTALL the Mac OS, restore everything from Time Machine, and restore the entire VM backup file

    Is anyone else seeing this?
    Is it a PD13 issue or is it an APFS issue?
    I think the only way to safely reclaim disk space is to quit out of everything else, then reclaim space.
    I have reclaimed disk space many times successfully but this has happened twice now.
    Any advice or observations would be much appreciated.
     
  2. JeremyRW

    JeremyRW Bit Poster

    Messages:
    5
    Yes, I can confirm the exact same thing happened to my machine.

    I ended up being able to fix my disk by deleting the parallels .pvm and all apfs local time machine snapshots using tmutil deletelocalsnapshots.

    After repeated restarts and time machine snapshot deletes I was able to get a clear fsck.
     
    Last edited: Jun 29, 2018
  3. Robert Swanson

    Robert Swanson Bit Poster

    Messages:
    39
    So I am NOT insane! My workaround is to disable automatic backups in Time Machine preferences. Just uncheck it, THEN shrink the VM file. I created another post with a WARNING. It's happened to me THREE times now. The third was when I forgot to disable auto backups.
     
  4. MarkS32

    MarkS32 Bit Poster

    Messages:
    4
    Hello Bob! I'm going through this problem now, First Aid can't help, same error as you had, root invalid. I had done an update to PD13 and it failed to launch. The next day I had a OSX update and then PD13 launched but Win8.1 wouldn't. I last had 160Gb free on a 500 Gb SSD but was down to 30. I've moved about 120 Gb of files to external drives yet I only have 62 Gb available. Then I did the First Aid and found the invalid fsroot. Oh, one thing that did speed me up was to reset PRAM, it dropped memory usage by 1.2 Gb after a purge command in terminal. Sorry I don't have more news. I've got a Time Machine backup and I'm hesitant to reload all the software, it's a pain to set up Windows with all my software. I'm looking for a fsroot fix without OSX/PD/Win reinstall.
     
  5. MarkS32

    MarkS32 Bit Poster

    Messages:
    4
    More notes: cannot copy documents/parallels/ pvm file to external drive for manual backup, process error dialog says cannot be read. Note the order of updates above, parallels 13 first, then it fails, the next day OS X update, parallels still fails and I discover the fsroot error and now can't copy pvm files. More later. I'm beginning to wonder if all would have been okay if OS X was updated first!
     
  6. Robert Swanson

    Robert Swanson Bit Poster

    Messages:
    39
    I haven't suffered through all that. I replaced my clunky old iMac desktop with this new one. The transfer of my data and applications was seamless. Moving PD13 was a piece of cake. It was incredibly simple. Until...I tried shrinking the VM file. EDIT: It is now THREE times. The last time was when I stupidly forgot to uncheck automatic backups. Sadly, I haven't heard one word from Parallels.
     
  7. MarkS32

    MarkS32 Bit Poster

    Messages:
    4
    Hello Bob!
    Thank you for your reply! My last name is Swanson and I have a cousin Bob! Any chance it's you? IL? Anyway, here's what has taken a week to figure out. First is the timing of the updates. A Parallels update before the OSX update caused the parallels pvm file to be uncopiyable! This makes me suspect the parallels update was not backwardly computable and brought the beta High Sierra fsroot problem to raise it's ugly head. My solution was to time machine backup, manual backup, then use time machine to bring back a 10 day old working copy of the pvm file. I then updated parallels again and updated windows 8.1. The system was stable again. It took 5 days to figure that out, 3 of them doing time machine backups and two manual backups because I don't trust any media and always back up manually regardless, to two drive. So, the parallels worked properly after the reimport and update of a 10 days old pvm file! The upshot for preventative maintenance is to always make sure OSX is updated before doing a parallels update, my best guess is that parallels did not make their update backwardly compatable thus the fsroot problem that caused the pvm file to be uncopiable. Any of this make sense to you? Thanks again. I finessed this problem by importing a 10 day old working pvm file and going through the virtual machine updates and all worked out fine. It's a structural problem with system software and I used an old technique to solve it!
    Cheers,
    Mark Swanson
     
  8. MarkS32

    MarkS32 Bit Poster

    Messages:
    4
    Once more a reminder, make sure OSX is updated before any Parallels update, that avoids any backward incompatibility in the Parallels update.
     
  9. Robert Swanson

    Robert Swanson Bit Poster

    Messages:
    39
    Hello everyone (anyone?).
    UPDATE: I thought I had a workaround by making a current backup, disabling automatic backups, then shrinking the VM. It doesn't seem to be the case. I started shrinking the VM yesterday evening and by this morning, it either had not completed (?!?) or was hung. I rebooted the Mac and, once again, my VM and file system was corrupted. Once again I had to perform a complete restore of my system. At least I know my Time Machine backups work very well! I am currently restoring my backup copy of my Win10 VM. Once it's finished, I will change the variable disk to fixed and be done with it. I'm tired of having to deal with this. It seems Parallels doesn't read this forum or they don't care. I think this is a serious issue between Parallels and High Sierra on APFS.
     
  10. StevensG

    StevensG

    Messages:
    1
    Same thing for myself. I lost my VM and my apfs was corrupter for the second time in 6 months. All the job done lost as I was committing in repository ... awesome and deceiving
     
  11. Robert Swanson

    Robert Swanson Bit Poster

    Messages:
    39
    I'm sorry to hear that. I changed my process (again) in order to (I hope) avoid this in the future. I shut down the VM, run a Time Machine backup (I do not back up the VM on Time Machine; it's separate), then turn OFF automatic backups, then reboot. I check 'About this Mac', 'Storage' to see if I have at least a decent amount of what IT considers free space, last, I shrink the VM file. I try to shut down all other apps, and find something else to do and walk away from the machine while it's reclaiming disk space. That may be overkill but I'm sick of wiping and restoring my Mac. 4 times now in 6 months. Sadly, I haven't heard a peep from anyone at Parallels on this forum about it.
     
  12. PatrickM16

    PatrickM16 Bit Poster

    Messages:
    1
    Thanks everyone for all of the information in this post, it was very helpful. Here are my notes, I hope they help someone else:

    I had this happen the other day where I believe Parallels locked up the system while shrinking a VM volume via "Real time virtual disk optimization". I ended up with these errors after I had to reboot the system:

    error: invalid dstream.size (89584566272), is greater than dstream.alloced_size (84972142592)
    error: xf : INO_EXT_TYPE_DSTREAM : invalid dstream
    error: inode_val: object (oid 0x4d872b): invalid xfields
    fsroot tree is invalid.

    The volume could not be verified completely.


    I have since disabled Optimization -> "Real time virtual disk optimization" on all of my VMs. I hope this ends the problem I have where after about 1 month of uptime, my VMs get slower, and I have problems shutting them down (or they just lock up on their own).

    To recover the VM file that was damage during the shrink, I was able to go into Time Machine, find the VM folder in the ~/Documents/Parallels folder, and restore a copy from time machine before the machine locked up to my Desktop, then delete the damaged files in ~/Documents/Parallels and move the files from my Desktop back to ~/Documents/Parallels.

    I was surprised I could use time machine to recover the VM since the Parallels folder is excluded from my time machine backups, but for whatever reason this worked (i'm guessing time machine restored it from the prior snapshot).

    Running "First Aid" in Disk Utility will still show problems with the snapshots, but deleting the bad APFS snapshots (see another post above) resolves that problem.
    Anyhow, I hope turning off the "Real time virtual disk optimization" does the trick in making these problems go away.
    Time will tell!
     
    Last edited: Aug 11, 2018
  13. Robert Swanson

    Robert Swanson Bit Poster

    Messages:
    39
    Very interesting. I've tried to read up on APFS and High Sierra. It seems users such as we have no control over these OS snapshots and how HS uses the hard drive space. I've had luck (so far) waiting until some Time Machine backups are completed and the snapshots go away. I wait until there is plenty of real free disk space and then shrink the VM HDD.
     

Share This Page