repairing a calendar

[Wolfgang Breyha]

Hi!

I'm currently debugging some oddities with a users default calendar after the
mailbox got XFER'ed to a different murder backend. The whole setup uses 2.5.11
currently.

After lightning started to fail the user removed the calendar and added it
again. lightning now tries to fetch the whole calendar (about 100 entries). I
see the REPORT request which gets a Multi-Status response with all ics URIs
and 200 OK. After that I see the calendar-multiget request for all these URIs.
But this request gets a HTTP/1.1 404 Not Found response. I assume that at
least one of the URIs fails. strace does not give any clues which one. The
session trace from cyrus doesn't help as well.

dumping the .dav sqlite DB it seems that there are at least 2 entries (the
latest ones after the XFER) which are not in the DB table and only in the
filesystem. The timestamp of the .dav DB is exactly from the XFER and not from
the newer entries found in the calendar folder.

Currently I'm searching my logs if there is something odd after the XFER
especially while the new entries got added.


While doing that the general questions raised how to repair a calendar at all,
since reconstruct does not do it as it seems? At least the .dav sqlite
database is not rebuilt. Removing the .dav file does no good as well. I get an
empty one after the next access to the calendar. But it is still not
accessible/syncable.

Greetings,
Wolfgang
-- 
Wolfgang Breyha <wbreyha at gmx.net> | http://www.blafasel.at/
Vienna University Computer Center | Austria