tdb_lock failed errors (bug #3327)

[Tony Mountifield]

In article <dpk09h$ji7$1 at softins.clara.co.uk>,
Tony Mountifield <tony at softins.clara.co.uk> wrote:
> Apologies for following up my own post, but I'm disappointed that despite
> people being active on the list, no-one has replied.....

Well, I don't know if this list is considered an exclusive clique or what,
but I never had any reply to my genuine attempt to investigate and fix a
valid bug. Poor show, really :-(

Thank you to coffeedude on IRC channel #samba-technical for taking to time
to discuss the bug with me.

For anyone who might find this thread when searching for info about the
same bug, I've now submitted a patch to fix the behaviour, which can be
found at https://bugzilla.samba.org/show_bug.cgi?id=3327

Hopefully one day the fix will make it into the code...

> I've done some more investigation, and I think this issue could be
> corrected in client/smbmount.c by including a call to namecache_shutdown()
> within the function send_fs_socket(), just inside the "if (!closed)"
> condition. I believe the next time the name cache is required it will
> automatically be opened again.

It turned out that not only was the call to namecache_shutdown() necessary,
but also a small correction to gencache_shutdown(), which wasn't clearing
out the "cache" pointer after closing the cache. This stopped it getting
re-opened properly when next needed.

Regards to all
Tony
-- 
Tony Mountifield
Work: tony at softins.co.uk - http://www.softins.co.uk
Play: tony at mountifield.org - http://tony.mountifield.org