Re: Any hangin lock issues, pserver, 1.11.10 & 1.11.11?

[Todd Denniston]

Mark wrote:
> 
> I have attempted to upgrade from 1.11.1p1 on solaris 2.8 to
> 1.11.11. We use a non-root pserver.
> 
> It seems that with 1.11.11(10), the 2 I have tryed, there quite
> often is lock files left, allong with the corresponding
> cvs-serv<id> directory. The server process exits, it is no longer
> running.
> 
> This lock files left are in pairs. What I mean is the #cvs.lock
> directory is left with the correspondind #cvs.wfl* file.
> 
> Any ideas... thought it was 1.11.11 so I when to 1.11.10.... when
> it still happened, I put the 1.11.1p1 binary back... everything
> is fine now.
> 
> I built the 1.11.1p1 with solaris make (ccs) and sun workshop
> 6.1, same with 1.11.11(10).... built them on the machine that
> they run on.
> 
> Thanks,
> 
> Mark
> 
Mark,
I am not sure if Sun got mmap correct in 2.8 (unpatched), but your problem
seems to have a boundary similar with a problem I had in solaris 2.6 recently.

http://mail.gnu.org/archive/html/bug-cvs/2003-11/msg00152.html

after you configure (but before compile) of cvs Version >1.11.1p1  make the
following change in config.h:

diff config.h.normal config.h
199c199
< #define HAVE_MMAP 1
---
> /*#define HAVE_MMAP 1*/


compile and try your hand tests with this version.

If this fixes your problem then it looks like you have a couple of options:
1) live with this no mmap version, and remember that you have to mod the
config.h file when putting new versions on because there is no --disable-mmap
option to configure.
2) patch the sun box up and see if that fixes it, with possible reversion to
option 1.

bonus option) slack off less than Todd and make a patch+doc_update to the
source adding a --disable-mmap config option and get Larry or Derek to apply
it.

-- 
Todd Denniston
Crane Division, Naval Surface Warfare Center (NSWC Crane) 
Harnessing the Power of Technology for the Warfighter