On 15/07/15 22:38, David Goodenough wrote: > I am running sid, so I have 3.8.10.2-1 and I am still getting these deadlocks. > What gives you the idea that they shoud have gone away?
Experimentation: I fed ~500 MiB maildir folder to akonadi/sqlite and
deadlocks disappeared with upgrade of sqlite3/libsqlite. It is quite
possible that I did not test long enough however I'm trying to determine
whether it is a real error or merely a warning?
Can you confirm?
By the way I think I've seen some quiet deadlocks (when akonadi doesn't
do anything while kmail waits for akonadi) even on backend-mysql. I
think akonadi _is_ a problem but I'm just not sure if deadlocks on
sqlite are really that bad, comparing to the whole thing.
In any case upstream admits that there are problems with concurrent
access on sqlite backend:
https://techbase.kde.org/Projects/PIM/Akonadi#Why_not_use_sqlite.3F
So as bad as it sounds sqlite just seems to be a poorly supported
backend... :(
--
Regards,
Dmitry Smirnov
signature.asc
Description: OpenPGP digital signature
