TDB on OpenBSD problem.

Andrew Tridgell tridge at samba.org
Wed May 9 06:27:35 GMT 2001


> No quotas on any of the file systems and no the file system was not full.
> It seemed like the record it was going to write was too big.

I wonder where the ENOSPC is coming from then? There is no code in tdb
that can return ENOSPC unless its being passed back from a system
call. But which system call, and why?

Can you check your user limits? On a sh like shell use "ulimit -a" and
on a csh like shell use "limit". Maybe there is some low local limit
we are hitting.

Do you have a system call tracer? It's called different things on
different OSes - common names are "strace", "ktrace/kdump", "trace"
and "par". I don't have a HPUX box so I don't know what it would be
called. A system call trace should tell us what syscall is failing.

We really are scraping the bottom of the barrel for ideas now. We
might need to wait for a systems programmer with access to a HPUX 11
box to have a deeper look at this. HP are currently trying to organise
a HPUX box for the team to use and that would probably help a lot, but
I imagine it will take a while before the box is up and running. (HPUX
is about the only major Unix-like OS missing from our build farm now -
see http://build.samba.org/)

Cheers, Tridge




More information about the samba-technical mailing list