I decided to separate a
snapshot-id (which already had several revisions) into two, representing two subsets of the same repository.
As I didn’t mind keeping the old revisions, but I didn’t want to upload again, I decided to:
- delete the old snapshot folder on storage
- perform a backup with the new snapshot 1
- perform a backup with the new snapshot 2
- run a
prune -exaustive -exclusiveto delete the old chunks
But in the end, I got the “missing chunks” error:
2. backup 1:
INFO BACKUP_STATS Total running time: 00:06:40 WARN BACKUP_SKIPPED 2 files were not included due to access errors
In fact there were two open files, but
check didn’t report anything wrong:
INFO SNAPSHOT_CHECK All chunks referenced by snapshot [new-snapshot-1] at revision 1 exist
3. backup 2:
INFO BACKUP_STATS Total running time: 00:22:02 WARN BACKUP_SKIPPED 6 files were not included due to access errors
INFO SNAPSHOT_CHECK All chunks referenced by snapshot [new-snapshot-2] at revision 1 exist FATAL DOWNLOAD_CHUNK Chunk f4580......14b017 can't be found
TRACE BACKBLAZE_CALL  URL request 'HEAD https://f001.backblazeb2.com/file/......14b017' returned status code 404 FATAL DOWNLOAD_CHUNK Chunk f4580......14b017 can't be found
I made sure the files were closed and ran the backup again, which ended without the “access errors” warnings, but the
check and the
prune continue to report the error.
In fact, the mentioned chunk is not in storage.
I removed the new created snapshot folders and executed the steps above again, with the same result. (and mentioning the same chunk!)