On Tue, 15 Mar 2005 15:29:16 -0500, Jan Harkes <jaharkes_at_cs.cmu.edu> wrote:
> Actually you are probably not placing them both on the same physical
> device or partition when you move one of them to a real dir.

duh! that was the case, indeed. Sorry.

> There is ofcourse a slight problem if currently chosen depth and width
> of the tree are not sufficient, changing either of those will make the
> server not find existing container files. I guess it would be useful to
> have something that can change the depth and width of the tree of
> container files in an existing partition, but we haven't actually had a
> case where we needed that.

That was what motivated me to create another partition. Well. i'm
simply removing everything and starting over those volumes. I now
created one with the 16M files option in vice-setupe-srvdir

But now i have a diferent error
i'm expanding a tar file into /coda and after some thousand files i
start receiving some instantaneous "time out" errors.

i stoped the client, did a venus -init and proceeded again acompaining
the log. Every time i do a "ls" in a dir with a few other dirs in it,
the venus log shows:
[ W(20) : 0000 : 11:30:15 ] Cachefile::SetLength 512

and the ls does its job.
But when i'm in the dir with the thousand files, the log shows:
[ W(20) : 0000 : 11:30:26 ] Cachefile::SetLength 165376

and then ls returns a lot of:
ls: 106955.php: Connection timed out
ls: 106744.php: Connection timed out
ad eternum

and if i don't cancel it soon venus dies
the server log has nothing besides a note about the new client.

# volutil rvmsize 1000003
V_BindToServer: binding to host camboinha
Volume 1000003 used a total of 1375132 bytes.
   8169 small vnodes used 914928 bytes.
   53 large vnodes used 54484 bytes.
    and 370688 bytes of DirPages.

Any other place that i can look for hints on what's going on?

Thanks again!
Gabriel