-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Daniel Önnerby wrote: > So what you are saying is that opening a SQLite DB on a shared network > drive SHOULD work with multiple clients (if all servers and NFS-version > are updated to most recent version)?
No, I am stating that the claim in the original message about oplocks breaking things is completely false. Oplocks have been in SMB since the mid 80s (yes really). They have been in NFS since v3 IIRC which would make it mid 90s. There are issues with other types of locking. I don't know of any in SMB implementations. I do know of some in NFS which was originally designed to be stateless and didn't use byte range locks. (Byte range locks are inherently stateful). Also due to poor coding, many NFS lock daemons would crash if the upper bits were set in a lock offset. (It isn't an error to lock beyond the end of the file. You can do poor man's IPC by using locks beyond the end of file.) UNIX was also never designed to do locking. I strongly recommend reading page 270 of the Unix Haters Guide. It also has an entire chapter about NFS. http://www.simson.net/ref/ugh.pdf If you are going to use a server that can be spoken to by two or more of NFS, SMB and local processes then you must either ensure oplocks are off, or that a kernel based oplock implementation is used. Linux, BSD and IRIX have had kernel oplocks since around 2000. Roger -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.6 (GNU/Linux) iD8DBQFFpf2TmOOfHg372QQRAm3BAJ9+vofCkP5slT3/aNiKZEd39pnBLACg4op3 F59UQLHdMP7FoKW1tY6syOU= =LjiP -----END PGP SIGNATURE----- ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
