Ghosts and starting over?

Support
#1

I just am having a heck of a time with the Check reporting GHOSTSNAPSHOTS and FOSIL size issues and I dont know what is all going on.

I want to start over… completely over I guess… What do I need to delete both on the server side (a NAS) and the client side (A Mac) so I can re-setup this backup and try again…

I just dont know what else to do

Thanks!!

#2
  1. Run check with fossils and persist
  2. Parse the log file and create a list of bad snapshots
  3. Go to your storage and delete those snapshots manually
  4. Prune exhaustive to get rid of orphans.
1 Like
#3

Thank you for this procedure. I am wondering… this is what I am seeing in my current prune log

INFO FOSSIL_GHOSTSNAPSHOT Snapshot savage1 revision 2 should have been deleted already
2022-01-10 11:11:56.770 INFO FOSSIL_GHOSTSNAPSHOT Snapshot savage1 revision 4 should have been deleted already
2022-01-10 11:11:56.770 INFO FOSSIL_GHOSTSNAPSHOT Snapshot savage1 revision 5 should have been deleted already
2022-01-10 11:11:56.770 INFO FOSSIL_GHOSTSNAPSHOT Snapshot savage1 revision 6 should have been deleted already

followed by
INFO SNAPSHOT_DELETE Deleting snapshot salvage1 at revision 2
2022-01-10 11:11:57.059 INFO SNAPSHOT_DELETE Deleting snapshot salvage1 at revision 4
2022-01-10 11:11:57.338 INFO SNAPSHOT_DELETE Deleting snapshot salvage1 at revision 5
2022-01-10 11:11:57.614 INFO SNAPSHOT_DELETE Deleting snapshot salvage1 at revision 6
etc

This is so freaking confusing… why does it say “should be deleted” followed by “deleting”?

This is why I feel I just want to start over… it seems like a mess…

Thoughts?

#4

It’s even more confusing. When it says “deleting” — it does not actually delete. It just marks it for deletion, and will delete later unless the prune is interrupted; in that case even more orphans will remain.

This is a long standing but, with multiple solutions suggested, hope it will eventually get fixed. In the meantime the workarounds are two:

  • never prune.
  • prune from a cloud host that is less likely to be interrupted.
#5

Thanks for your expertise!!

Is it ok for me to go to the storage and delete all or maybe all but the latest snapshot? Just manually delete them? Is that ok technically speaking? I don’t need old data at the moment… just looking for the simplest way to get it to run clean given that I don’t need old snapshots…

#6

Yes, you can delete the snapshot files form “snapshots” folder, and then run prune -exhaustive to delete now-unreferenced chunks and free some storage.

#7

What might be a reason to not run -exhaustive on all prune operations?

#8

-exhaustive will take considerably longer time.

Regular prune figures out which snapshots to delete based on specified retention policy, and therefore the list of potential chunks that need to be considered for deletion is limited to what the candidate snapshots are referring to.

-exhaustive prune checks all chunks in the datastore to identify those that are not referecened by any snapshot. Therefore it has to analyze all snapshots, all revision, and get list of all chunks.

#9

Got it… thanks again!