Hi,

> [...] It's a mount/rsync/dismount; I'm pretty sure the unmount moves
> the metadata. I'd prefer every 24 hours, if that could be managed, but
> they are/were every hour.[...]
that's correct.

> [...]
> I still have a snapshot of the old bad database I think...is there a
> query you could suggest to find the offending record, that might jog
> my memory.
> I tried:
> select * from contents c join contents c1 on c1.parent_inode =
> c.parent_inode and c1.name_id = c.name_id and c1.rowid > c.rowid;
> and got a lot of rows...which is a tad scary given the constraint on
> the table.
That SQL command looks ok to me. Maybe also try to run a "REINDEX
contents;" (https://sqlite.org/lang_reindex.html)
Also check with ".schema contents" and ".indexes contents" if that index
does exist.
(https://sqlite.org/cli.html#special_commands_to_sqlite3_dot_commands_)

Try to run the same command on the now fixed database (making a copy to
be on the safe side). It should not show any results. If that is the
case, then you somehow got many duplicate files in directories. Are the
parent_inodes all the same (problem was only in one folder) or different?

-- 
You received this message because you are subscribed to the Google Groups 
"s3ql" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/s3ql/90b46fcd-760e-b43b-5f8d-4e9d2f57d7a6%40jagszent.de.

Reply via email to