[ZODB-Dev] Berkeley Storage Checkpointing

[Toby Dickenson]

Setting up my production BerkeleyStorage system, it appears that one
remaining uglyness is the need for checkpointing.

Sleepycat's db_checkpoint daemon helps a little, although it is still
ugly. Also, I suspect bad things would happen if checkpointing kicked
in while the storage was in the recovery phase of opening the
database.

How does anyone else handle checkpointing?

Barry, would you accept a patch to BerkeleyBase.py to start a separate
thread for regular checkpointing?

Toby Dickenson
tdickenson@geminidataloggers.com